chore: release v0.4.0 (#124)

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

.claude/commands/fix-issue.md

@@ -0,0 +1,97 @@
+---
+description: Fetch a GitHub issue, create a branch, research the codebase, plan the fix, implement with tests, and commit
+disable-model-invocation: true
+allowed-tools: Bash(gh issue view:*), Bash(gh repo view:*), Bash(git fetch:*), Bash(git checkout:*), Bash(git status:*), Bash(git branch:*), Bash(git add:*), Bash(git commit:*), Bash(cargo fmt:*), Bash(cargo clippy:*), Bash(cargo test:*), Read, Edit, Write, Grep, Glob
+argument-hint: "<issue-number or github-issue-url>"
+---
+
+# Fix GitHub Issue
+
+## Step 1: Resolve the issue
+
+Parse `$ARGUMENTS` to extract the issue number:
+- If it's a URL like `https://github.com/owner/repo/issues/42`, extract `42`.
+- If it's a bare number, use it directly.
+- If empty, stop and ask the user for an issue number.
+
+Fetch the issue:
+
+```
+gh issue view {number} --json title,body,labels,assignees,comments,state
+```
+
+If the issue is closed, warn the user and ask if they still want to proceed.
+
+## Step 2: Create a branch
+
+Create a fresh branch off the latest main:
+
+1. Fetch latest: `git fetch origin`
+2. Detect default branch: `gh repo view --json defaultBranchRef --jq .defaultBranchRef.name`
+3. Create and switch to a new branch: `git checkout -b fix/{number}-{short-slug} origin/{default-branch}`
+   - `{short-slug}` is 3-5 words from the issue title, lowercase, hyphenated (e.g. `fix/42-idor-workspace-check`)
+
+If the working tree has uncommitted changes, warn the user and stop. Do not stash or discard their work.
+
+## Step 3: Understand the issue
+
+Summarize the issue in 2-3 sentences. Identify:
+- **What's broken or missing** (the symptom or feature request)
+- **Acceptance criteria** (what "done" looks like, from the issue body or comments)
+- **Constraints** (mentioned technologies, backward compatibility, performance requirements)
+
+If the issue is unclear or ambiguous, list the open questions. These will be addressed during planning.
+
+## Step 4: Research the codebase
+
+Before planning, gather context:
+
+1. **Find relevant code** - Search for files, functions, types, and patterns mentioned in the issue. Read them in full.
+2. **Trace the flow** - If the issue is about a specific behavior, trace the code path from the entry point (route handler, CLI command, etc.) through to the relevant logic.
+3. **Check existing tests** - Find tests related to the affected code. Understand what's already covered.
+4. **Check for prior art** - Look for similar patterns in the codebase that solve analogous problems. Prefer consistency with existing patterns.
+
+## Step 5: Enter planning mode
+
+Enter planning mode to design the implementation. The plan MUST cover:
+
+1. **Root cause** (for bugs) or **design approach** (for features)
+2. **Files to modify** with specific descriptions of what changes in each
+3. **New files** (if any) with justification for why they're needed
+4. **Tests to add** - every code path introduced or changed needs a test:
+   - Happy path (expected input produces expected output)
+   - Error paths (invalid input, missing data, permission denied)
+   - Edge cases (empty collections, boundary values, concurrent access)
+5. **IronClaw-specific concerns**:
+   - If the change touches persistence, both database backends must be updated (`postgres.rs` and `libsql_backend.rs`)
+   - New `Database` trait methods need implementations in both backends
+   - No `.unwrap()` or `.expect()` in production code
+   - Use `crate::` imports, not `super::`
+   - Error types via `thiserror` in `error.rs`
+6. **Migration or compatibility concerns** (if any)
+
+Follow the project's CLAUDE.md guidance for architecture decisions.
+
+Wait for user approval before implementing.
+
+## Step 6: Implement
+
+After the plan is approved:
+
+1. Implement each change from the plan.
+2. Write all planned tests.
+3. Run IronClaw's full quality gate:
+   - `cargo fmt`
+   - `cargo clippy --all --benches --tests --examples --all-features` (zero warnings)
+   - `cargo test --lib` (all tests pass)
+4. If any check fails, fix it before proceeding.
+
+Note: Integration tests (`--test workspace_integration`) require PostgreSQL and are expected to fail locally. Only `--lib` test failures are blocking.
+
+## Step 7: Commit and summarize
+
+1. Commit with a descriptive message referencing the issue (e.g. `fix: prevent IDOR in function call outputs (#42)`).
+2. Summarize what was done:
+   - Files changed with line references
+   - Tests added and what they cover
+   - Any follow-up work or open questions

.claude/commands/respond-pr.md

@@ -0,0 +1,81 @@
+---
+description: Respond to PR review comments — triage, plan fixes, implement after confirmation, push, and reply to reviewers
+disable-model-invocation: true
+allowed-tools: Bash(gh pr list:*), Bash(gh pr comment:*), Bash(gh api:*), Bash(gh repo view:*), Bash(git branch:*), Bash(git status:*), Bash(git add:*), Bash(git commit:*), Bash(git push:*), Bash(cargo fmt:*), Bash(cargo clippy:*), Bash(cargo test:*), Read, Edit, Write, Grep, Glob
+argument-hint: "[pr-number (optional, auto-detects from branch)]"
+---
+
+# Review and Address PR Comments
+
+## Step 1: Find the PR
+
+If `$ARGUMENTS` is provided, use that as the PR number. Otherwise, detect the PR for the current branch:
+
+```
+gh pr list --head $(git branch --show-current) --json number,title,url --jq '.[0]'
+```
+
+If no PR is found, tell the user and stop.
+
+## Step 2: Fetch all review comments
+
+Resolve the repo owner and name:
+
+```
+gh repo view --json owner,name --jq '"\(.owner.login)/\(.name)"'
+```
+
+Fetch the full set of review comments (not issue-level comments):
+
+```
+gh api --paginate repos/{owner}/{repo}/pulls/{number}/comments
+```
+
+Also fetch the review summaries:
+
+```
+gh api --paginate repos/{owner}/{repo}/pulls/{number}/reviews
+```
+
+Deduplicate comments that appear multiple times (bots sometimes post the same finding under different IDs). Group by the actual issue being raised, not by comment ID.
+
+## Step 3: Triage and plan
+
+For each unique issue raised in the comments:
+
+1. **Check if already addressed** - Read the current code at the referenced location. If a prior commit already fixed it, note it as "already resolved".
+2. **Assess validity** - Determine if the comment identifies a real problem or is a false positive. Be honest about false positives but explain why.
+3. **Classify severity** - Critical (security/data loss), High (bugs/broken behavior), Medium (correctness/robustness), Low (style/naming/nits).
+4. **Plan the fix** - For each valid unresolved issue, describe the specific code change needed.
+
+Present the plan as a table to the user:
+
+| # | Issue | File:Line | Severity | Status | Planned Fix |
+|---|-------|-----------|----------|--------|-------------|
+
+Wait for user confirmation before proceeding to implementation.
+
+## Step 4: Implement fixes
+
+After user confirms:
+
+1. Implement each fix in the plan.
+2. Run IronClaw's quality gate to verify nothing breaks:
+   - `cargo fmt`
+   - `cargo clippy --all --benches --tests --examples --all-features`
+   - `cargo test --lib`
+3. Commit with a descriptive message referencing the PR review.
+4. Push to the branch.
+
+## Step 5: Reply to comments
+
+For each comment addressed, reply on the PR with a short message stating what was fixed and the commit SHA. For false positives or already-resolved items, reply explaining why no change was needed.
+
+## Rules
+
+- Never guess at code you haven't read. Always read the referenced file and line before assessing a comment.
+- Group duplicate comments (same issue reported by multiple bots) and reply to all of them.
+- Do not make changes beyond what the review comments ask for. Stay focused.
+- If a comment suggests a change you disagree with, present your reasoning to the user during the planning phase rather than silently ignoring it.
+- Follow IronClaw conventions: no `.unwrap()` in production code, use `crate::` imports, `thiserror` errors.
+- If changes touch persistence, verify both database backends are updated.

.claude/commands/review-crate.md

@@ -0,0 +1,245 @@
+---
+description: Deep audit of the IronClaw crate for vulnerabilities, bugs, unfinished work, inconsistencies, and oversights
+disable-model-invocation: true
+allowed-tools: Bash(cargo fmt:*), Bash(cargo clippy:*), Bash(cargo test:*), Bash(cargo audit:*), Bash(git diff:*), Bash(git log:*), Bash(git show:*), Bash(wc:*), Read, Grep, Glob, Task
+argument-hint: "[path/to/crate]"
+---
+
+# Rust Crate Audit
+
+You are performing a thorough audit of a Rust crate. Your goal is to find every vulnerability, bug, unfinished piece of work, inconsistency, and oversight before it ships. Leave no stone unturned.
+
+## Step 1: Locate the crate
+
+Parse `$ARGUMENTS`:
+- If a path is provided, use it as the crate root.
+- If empty, use the current working directory.
+
+Verify it's a valid Rust crate by checking for `Cargo.toml`. If not found, stop and ask the user.
+
+## Step 2: Understand the crate
+
+Read `Cargo.toml` to understand:
+- Crate name, version, edition
+- Dependencies (look for outdated, unmaintained, or suspicious crates)
+- Feature flags and their implications
+- Build scripts (`build.rs`) if any
+
+Read `CLAUDE.md`, `README.md`, or top-level documentation if present to understand intent and architecture.
+
+Read `src/lib.rs` or `src/main.rs` to get the module tree. Then read each module's `mod.rs` or top-level file to build a mental map of the crate's structure before diving into details.
+Read all Rust files (`src/*.rs`) to make sure everything is in context when you are reasoning.
+
+## Step 3: Run the compiler's checks
+
+Run these commands and capture output. Do NOT fix anything, just collect findings:
+
+```
+cargo fmt --check 2>&1
+```
+
+```
+cargo clippy --all --benches --tests --examples --all-features -- -W clippy::all -W clippy::pedantic -W clippy::nursery 2>&1
+```
+
+```
+cargo test --lib 2>&1
+```
+
+If any of these fail, record the failures as findings. If `cargo test` has ignored tests, note which ones and why.
+
+Note: Integration tests (`--test workspace_integration`) require a PostgreSQL database and are expected to fail locally. Only report `--lib` test failures as blocking.
+
+## Step 4: Scan for unfinished work
+
+Search the entire `src/` tree for:
+
+```
+todo!
+unimplemented!
+fixme
+FIXME
+TODO
+HACK
+XXX
+SAFETY:
+stub
+placeholder
+temporary
+```
+
+For each match:
+- Is it in production code or test code?
+- Is it a genuine incomplete feature or a deliberate placeholder?
+- Is there a tracking issue referenced?
+- Could this panic at runtime?
+
+Any `todo!()` or `unimplemented!()` in non-test code is **High severity** (runtime panic).
+
+## Step 5: Audit for vulnerabilities and unsafe code
+
+### 5a. Unsafe code
+
+Search for all `unsafe` blocks. For each one:
+- Is the safety invariant documented with a `// SAFETY:` comment?
+- Is the invariant actually upheld by the surrounding code?
+- Could the unsafe block be replaced with a safe alternative?
+- Are there any pointer dereferences, transmutes, or FFI calls?
+
+### 5b. Unwrap and panic paths
+
+Search for `.unwrap()`, `.expect(`, `panic!`, `unreachable!` in non-test code. For each:
+- Can this actually panic in production?
+- Is there a code path that reaches this with None/Err?
+- Should it be replaced with proper error handling (`?`, `.ok()`, `.unwrap_or_default()`)?
+
+IronClaw convention: `.unwrap()` and `.expect()` are banned in production code. Any occurrence outside `#[cfg(test)]` blocks is a **High severity** finding.
+
+### 5c. SQL and injection vectors
+
+Search for string formatting used in SQL queries, shell commands, or HTML:
+- `format!` used near `.execute(`, `.query(`, `Command::new(`
+- String interpolation in query construction vs parameterized queries
+- User input flowing into file paths (`Path::new`, `std::fs::`)
+
+IronClaw has two database backends (PostgreSQL and libSQL). Check both for injection vectors.
+
+### 5d. Cryptographic issues
+
+If the crate uses crypto:
+- Are comparisons constant-time? (look for `==` on secrets/hashes vs `subtle::ConstantTimeEq`)
+- Is randomness from `OsRng` / `thread_rng` and not a fixed seed?
+- Are keys/secrets zeroized after use? (`secrecy`, `zeroize` crates)
+- Are deprecated algorithms used? (MD5, SHA1 for security, RC4, DES)
+
+### 5e. Resource exhaustion
+
+- Are there unbounded allocations? (`Vec` growing from user input without limits)
+- Are there unbounded loops? (retry loops without max attempts)
+- Are file reads bounded? (`std::fs::read_to_string` on user-provided paths)
+- Are timeouts set on all network operations?
+- Are there connection/resource leaks? (opened but never closed, missing `Drop`)
+
+### 5f. Error handling
+
+- Are errors swallowed silently? (`let _ = ...`, `.ok()` discarding errors that matter)
+- Do error types carry enough context to debug in production?
+- Are there error type mismatches? (returning generic `anyhow::Error` where a typed error would prevent confusion)
+- Is `thiserror` used consistently for error types (IronClaw convention)?
+
+## Step 6: Check for inconsistencies
+
+### 6a. Naming conventions
+
+- Are types, functions, modules named consistently? (e.g., mixing `get_` and `fetch_`, `create_` and `new_`)
+- Do similar operations follow the same patterns?
+
+### 6b. Duplicate or near-duplicate code
+
+Look for:
+- Functions that do nearly the same thing with minor variations (candidates for generics or shared helpers)
+- Repeated error mapping patterns that should be extracted
+- Copy-pasted SQL queries or string templates with slight differences
+- Identical struct definitions or conversion logic in different modules
+
+### 6c. API consistency
+
+- Do similar functions take arguments in the same order?
+- Are return types consistent? (e.g., some functions return `Option<T>`, similar ones return `Result<T, E>`)
+- Are visibility modifiers consistent? (`pub` where it should be `pub(crate)`, or vice versa)
+
+### 6d. Dead code and unused items
+
+- Are there functions, structs, or modules that nothing references?
+- Are there `#[allow(dead_code)]` annotations that should be investigated?
+- Are there feature-gated items where the feature is never enabled?
+
+### 6e. Import style
+
+IronClaw convention: use `crate::` imports, not `super::`. Flag any `super::` imports in non-test code.
+
+## Step 7: Inspect for change oversights
+
+### 7a. Partial refactors
+
+- Are there old patterns coexisting with new patterns?
+- Are there renamed types/functions where some call sites still use the old name via a compatibility alias?
+- Are there comments referencing behavior that no longer exists?
+
+### 7b. Trait implementation gaps
+
+- If a trait is defined, do all intended types implement it?
+- Are there `impl` blocks that look incomplete?
+- Are `Default` implementations sensible?
+
+IronClaw key traits: `Database` (~60 methods), `Channel`, `Tool`, `LlmProvider`, `SuccessEvaluator`, `EmbeddingProvider`. If any new methods were added to `Database`, verify both `postgres.rs` and `libsql_backend.rs` implement them.
+
+### 7c. Test coverage gaps
+
+- Are there public functions without any test?
+- Are there error paths without tests?
+- Are there recently-changed functions where the tests still assert old behavior?
+
+### 7d. Documentation drift
+
+- Do doc comments match actual function behavior?
+- Are examples in doc comments still valid and compilable?
+
+## Step 8: Dependency audit
+
+Review `Cargo.toml` and `Cargo.lock`:
+- Are there duplicate versions of the same crate in the lock file? (potential version conflicts)
+- Are there dependencies with known security advisories? Run `cargo audit` to check (install with `cargo install cargo-audit` if not present).
+- Are there heavy dependencies used for trivial functionality?
+- Are dependency features minimal?
+
+## Step 9: Present findings
+
+Compile all findings into a structured report. Group by severity, then by category.
+
+### Format
+
+For each finding:
+
+```
+### [Severity] Category: One-line summary
+
+**Location:** `file_path:line_number`
+**Category:** Vulnerability | Bug | Unfinished | Inconsistency | Duplicate | Oversight | Style
+
+**Description:**
+Detailed explanation of the issue, why it matters, and how it could manifest.
+
+**Suggested fix:**
+Concrete suggestion with code if applicable.
+```
+
+### Severity levels
+
+- **Critical**: Security vulnerability, data loss, or crash in production
+- **High**: Bug that causes incorrect behavior, `todo!()`/`unimplemented!()` in prod code, or missing validation on trust boundaries
+- **Medium**: Inconsistency, duplicate code, incomplete error handling, missing tests for important paths
+- **Low**: Naming inconsistency, unnecessary complexity, documentation drift, minor dead code
+- **Nit**: Style preference, optional improvement
+
+### Summary table
+
+End with a summary table:
+
+| # | Severity | Category | File:Line | Finding |
+|---|----------|----------|-----------|---------|
+
+And a final tally: X Critical, Y High, Z Medium, W Low, V Nit.
+
+## Rules
+
+- Read every file before reporting on it. Never guess about code you haven't seen.
+- Be specific. "This might have issues" is worthless. "Line 42 calls `.unwrap()` on a `Result` that returns `Err` when the DB connection is dropped" is useful.
+- Distinguish certainty levels: "this IS a bug" vs "this COULD be a bug if X".
+- Don't invent problems to look thorough. If the code is solid, say so.
+- Focus on substance over style. Don't flag formatting unless it causes real confusion.
+- Respect existing project conventions (check CLAUDE.md). Don't flag patterns the project explicitly endorses.
+- When in doubt about severity, round up.
+- For large crates (>50 files), prioritize: core logic > public API > internal utilities > tests > examples.
+- Use the Task tool to parallelize file reading across modules when the crate is large.
+- Do NOT fix anything. This is a read-only audit. Report findings for the user to action.

.claude/commands/review-pr.md

@@ -0,0 +1,170 @@
+---
+description: Paranoid architect review of a PR — fetches diff, reads changed files, deep review across 6 lenses, posts findings as GitHub comments
+disable-model-invocation: true
+allowed-tools: Bash(gh pr view:*), Bash(gh pr diff:*), Bash(gh pr comment:*), Bash(gh api:*), Bash(gh repo view:*), Bash(git diff:*), Bash(git log:*), Read, Grep, Glob
+argument-hint: "<pr-number or github-pr-url>"
+---
+
+# Paranoid Architect Code Review
+
+You are reviewing this PR as a paranoid architect. Your job is to find every bug, vulnerability, race condition, edge case, and undocumented assumption before it ships. Assume adversarial users, concurrent access, and Murphy's law.
+
+## Step 1: Resolve the PR
+
+Parse `$ARGUMENTS` to extract the PR number:
+- If it's a URL like `https://github.com/owner/repo/pull/123`, extract `123`.
+- If it's a bare number, use it directly.
+- If empty, stop and ask the user for a PR number.
+
+Fetch PR metadata (including head commit SHA for posting line comments later):
+
+```
+gh pr view {number} --json title,body,baseRefName,headRefName,headRefOid,files,additions,deletions
+```
+
+Save the `headRefOid` value, you'll need it as `commit_id` in Step 6.
+
+## Step 2: Load the full diff
+
+```
+gh pr diff {number}
+```
+
+Also get the list of changed files:
+
+```
+gh pr diff {number} --name-only
+```
+
+## Step 3: Read every changed file in full
+
+For each changed file, read the ENTIRE current file (not just the diff hunks). You need surrounding context to catch:
+- Callers of modified functions that now behave differently
+- Trait/interface contracts that the change may violate
+- Invariants established elsewhere that the diff breaks
+
+If the PR touches more than 20 files, still read all of them, but process in this priority order: service logic > routes/handlers > models/types > tests > docs. Batch reads in groups of ~20 if needed.
+
+## Step 4: Deep review
+
+Go through the changes with each of these lenses. For every finding, note the file, line range, severity, and a concrete description.
+
+### IronClaw-specific checks
+
+In addition to the general lenses below, check IronClaw conventions (see CLAUDE.md):
+- No `.unwrap()` or `.expect()` in production code (tests are fine)
+- Use `crate::` imports, not `super::`
+- Error types use `thiserror` in `error.rs`
+- If the change touches persistence, verify both database backends are updated (PostgreSQL in `postgres.rs` AND libSQL in `libsql_backend.rs`)
+- New tools must implement the `Tool` trait correctly and be registered in `registry.rs`
+- External tool output must pass through the safety layer
+
+### 4a. Correctness and bugs
+
+- Off-by-one errors, wrong comparison operators, inverted conditions
+- Unreachable code, dead branches, impossible match arms
+- Type confusion (mixing up IDs, using wrong enum variant)
+- Incorrect error propagation (swallowed errors, wrong error type/status code)
+- Broken invariants (e.g. uniqueness assumptions violated, ordering assumptions wrong)
+- Concurrency issues (TOCTOU, missing locks, race conditions between check and use)
+
+### 4b. Edge cases and failure handling
+
+- What happens with empty input, None/null, zero-length collections?
+- What happens when external services fail (DB down, HTTP timeout, malformed response)?
+- What happens at integer boundaries (overflow, underflow, i64::MAX)?
+- What happens with malformed or adversarial input (invalid UTF-8, huge payloads, deeply nested JSON)?
+- Are all error paths tested? Does every `?` propagation make sense?
+- Are partial failures handled (e.g. wrote to DB but failed to emit event)?
+
+### 4c. Security (assume a malicious actor)
+
+- **Authentication/Authorization bypass**: Can an unauthenticated user reach this? Can workspace A's user access workspace B's data? Are there IDOR vulnerabilities?
+- **Injection**: SQL injection via string interpolation? Command injection? Log injection? Header injection?
+- **Data leakage**: Are secrets, PII, or conversation content logged? Returned in error messages? Exposed in API responses?
+- **Resource exhaustion / DoS**: Can an attacker send unbounded input? Trigger expensive operations without rate limits? Cause OOM via large allocations?
+- **Financial abuse**: Can tokens/credits be consumed without being tracked? Can usage limits be bypassed?
+- **Replay / race conditions**: Can the same request be replayed for double-spend? Can concurrent requests bypass limits?
+- **Cryptographic issues**: Timing attacks on comparisons? Weak randomness? Missing HMAC verification?
+
+### 4d. Test coverage
+
+- Is every new public function/method tested?
+- Are error paths tested (not just happy paths)?
+- Are edge cases covered (empty input, boundary values, concurrent access)?
+- Do existing tests still make sense with the new changes, or do they assert stale behavior?
+- Are there integration/e2e tests for the full flow?
+- If a test is missing, describe exactly what test should be written.
+
+### 4e. Documentation and assumptions
+
+- Are new assumptions documented in comments? (e.g. "this field is always non-empty because X")
+- Are non-obvious algorithms or business rules explained?
+- Are API contracts (request/response shapes, error codes, status codes) documented?
+- Are there TODO/FIXME/HACK comments that should be tracked as issues?
+
+### 4f. Architectural concerns
+
+- Does this change follow existing patterns in the codebase, or does it introduce a new one without justification?
+- Are there unnecessary abstractions or premature generalizations?
+- Is there duplicated logic that should be extracted?
+- Are dependencies between modules clean, or does this create circular/tight coupling?
+- Will this change make future work harder?
+
+## Step 5: Present findings
+
+Summarize findings to the user as a table:
+
+| # | Severity | Category | File:Line | Finding | Suggested Fix |
+|---|----------|----------|-----------|---------|---------------|
+
+Severity levels:
+- **Critical**: Security vulnerability, data loss, or financial exploit
+- **High**: Bug that will cause incorrect behavior in production
+- **Medium**: Robustness issue, missing validation, or incomplete error handling
+- **Low**: Style, naming, documentation, or minor improvement
+- **Nit**: Optional suggestion, take-it-or-leave-it
+
+Ask the user which findings to post as PR comments. Default: all Critical, High, and Medium.
+
+## Step 6: Post comments on GitHub
+
+Resolve the repo owner and name if not already known:
+
+```
+gh repo view --json owner,name --jq '"\(.owner.login)/\(.name)"'
+```
+
+For each approved finding, post a review comment on the PR at the specific file and line. Use the `headRefOid` from Step 1 as the `commit_id`:
+
+```
+gh api repos/{owner}/{repo}/pulls/{number}/comments \
+  -f body="..." \
+  -f path="..." \
+  -f commit_id="{headRefOid}" \
+  -F line=... \
+  -f side="RIGHT"
+```
+
+For findings that span multiple locations or are architectural, post as a regular PR comment:
+
+```
+gh pr comment {number} --body "..."
+```
+
+Format each comment clearly:
+- Severity tag (e.g. `**High Severity**`)
+- One-line summary
+- Detailed explanation of the issue
+- Concrete suggestion for the fix (with code if possible)
+
+## Rules
+
+- Read every changed file in full before writing a single finding. Context matters.
+- Never post a comment about code you haven't actually read. Verify line numbers against the actual file.
+- Be specific. "This might have issues" is useless. "Line 42 returns 404 but should return 400 because X" is useful.
+- Distinguish between "this IS a bug" and "this COULD be a bug if X". Be honest about certainty.
+- Don't nitpick formatting or style unless it causes actual confusion. Focus on substance.
+- If the code is good and you find nothing, say so. Don't invent problems to look thorough.
+- Respect the project's CLAUDE.md privacy rules: never include customer data, secrets, or PII in comments.
+- When in doubt about severity, round up. It's cheaper to dismiss a false alarm than to miss a real bug.

.github/workflows/code_style.yml

@@ -8,12 +8,13 @@ jobs:
     runs-on: ubuntu-latest
     steps:
     - name: Checkout repository
-      uses: actions/checkout@v4
+      uses: actions/checkout@v6
     - name: Install Rust
       uses: dtolnay/rust-toolchain@stable
       with:
         profile: minimal
         components: rustfmt, clippy
+    - uses: Swatinem/rust-cache@v2
     - name: Check formatting
       run: |
         cargo fmt --all -- --check

.github/workflows/release-plz.yml

@@ -24,6 +24,7 @@ jobs:
       - &install-rust
         name: Install Rust toolchain
         uses: dtolnay/rust-toolchain@stable
+      - uses: Swatinem/rust-cache@v2
       # Generating a GitHub token, so that PRs and tags created by
       # the release-plz-action can trigger actions workflows.
       - name: Generate GitHub token
@@ -56,6 +57,7 @@ jobs:
     steps:
       - *checkout
       - *install-rust
+      - uses: Swatinem/rust-cache@v2
       - name: Run release-plz
         uses: release-plz/[email protected]
         with:

.github/workflows/release.yml

@@ -39,7 +39,6 @@ permissions:
 # If there's a prerelease-style suffix to the version, then the release(s)
 # will be marked as a prerelease.
 on:
-  pull_request:
   push:
     tags:
       - '**[0-9]+.[0-9]+.[0-9]+*'
@@ -282,43 +281,14 @@ jobs:
 
           gh release create "${{ needs.plan.outputs.tag }}" --target "$RELEASE_COMMIT" $PRERELEASE_FLAG --title "$ANNOUNCEMENT_TITLE" --notes-file "$RUNNER_TEMP/notes.txt" artifacts/*
 
-  publish-npm:
-    needs:
-      - plan
-      - host
-    runs-on: "ubuntu-22.04"
-    env:
-      GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      PLAN: ${{ needs.plan.outputs.val }}
-    if: ${{ !fromJson(needs.plan.outputs.val).announcement_is_prerelease || fromJson(needs.plan.outputs.val).publish_prereleases }}
-    steps:
-      - name: Fetch npm packages
-        uses: actions/download-artifact@v4
-        with:
-          pattern: artifacts-*
-          path: npm/
-          merge-multiple: true
-      - uses: actions/setup-node@v4
-        with:
-          node-version: '20.x'
-          registry-url: 'https://registry.npmjs.org'
-      - run: |
-          for release in $(echo "$PLAN" | jq --compact-output '.releases[] | select([.artifacts[] | endswith("-npm-package.tar.gz")] | any)'); do
-            pkg=$(echo "$release" | jq '.artifacts[] | select(endswith("-npm-package.tar.gz"))' --raw-output)
-            npm publish --access public "./npm/${pkg}"
-          done
-        env:
-          NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-
   announce:
     needs:
       - plan
       - host
-      - publish-npm
     # use "always() && ..." to allow us to wait for all publish jobs while
     # still allowing individual publish jobs to skip themselves (for prereleases).
     # "host" however must run to completion, no skipping allowed!
-    if: ${{ always() && needs.host.result == 'success' && (needs.publish-npm.result == 'skipped' || needs.publish-npm.result == 'success') }}
+    if: ${{ always() && needs.host.result == 'success' }}
     runs-on: "ubuntu-22.04"
     env:
       GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/test.yml

@@ -11,10 +11,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
         with:
           profile: minimal
+      - uses: Swatinem/rust-cache@v2
       - name: Run Tests
         run: cargo test --all-features -- --nocapture

.gitignore

@@ -1,6 +1,15 @@
 
 .env
 .env.local
+.env.*
+!.env.example
+
+# Claude Code worktrees
+.claude/worktrees/
+
+# Sidecar tool data
+.sidecar/
+.todos/
 
 target/
 

CHANGELOG.md

@@ -7,6 +7,67 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.5.0](https://github.com/nearai/ironclaw/compare/v0.4.0...v0.5.0) - 2026-02-17
+
+### Added
+
+- add cooldown management to FailoverProvider ([#114](https://github.com/nearai/ironclaw/pull/114))
+
+## [0.4.0](https://github.com/nearai/ironclaw/compare/v0.3.0...v0.4.0) - 2026-02-17
+
+### Added
+
+- move per-invocation approval check into Tool trait ([#119](https://github.com/nearai/ironclaw/pull/119))
+- add polished boot screen on CLI startup ([#118](https://github.com/nearai/ironclaw/pull/118))
+- Add lifecycle hooks system with 6 interception points ([#18](https://github.com/nearai/ironclaw/pull/18))
+
+### Other
+
+- remove accidentally committed .sidecar and .todos directories ([#123](https://github.com/nearai/ironclaw/pull/123))
+
+## [0.3.0](https://github.com/nearai/ironclaw/compare/v0.2.0...v0.3.0) - 2026-02-17
+
+### Added
+
+- direct api key and cheap model ([#116](https://github.com/nearai/ironclaw/pull/116))
+
+## [0.2.0](https://github.com/nearai/ironclaw/compare/v0.1.3...v0.2.0) - 2026-02-16
+
+### Added
+
+- mark Ollama + OpenAI-compatible as implemented ([#102](https://github.com/nearai/ironclaw/pull/102))
+- multi-provider inference + libSQL onboarding selection ([#92](https://github.com/nearai/ironclaw/pull/92))
+- add multi-provider LLM failover with retry backoff ([#28](https://github.com/nearai/ironclaw/pull/28))
+- add libSQL/Turso embedded database backend ([#47](https://github.com/nearai/ironclaw/pull/47))
+- Move debug log truncation from agent loop to REPL channel ([#65](https://github.com/nearai/ironclaw/pull/65))
+
+### Fixed
+
+- shell destructive-command check bypassed by Value::Object arguments ([#72](https://github.com/nearai/ironclaw/pull/72))
+- propagate real tool_call_id instead of hardcoded placeholder ([#73](https://github.com/nearai/ironclaw/pull/73))
+- Fix wasm tool schemas and runtime ([#42](https://github.com/nearai/ironclaw/pull/42))
+- flatten tool messages for NEAR AI cloud-api compatibility ([#41](https://github.com/nearai/ironclaw/pull/41))
+- security hardening across all layers ([#35](https://github.com/nearai/ironclaw/pull/35))
+
+### Other
+
+- Explicitly enable cargo-dist caching for binary artifacts building
+- Skip building binary artifacts on every PR
+- add module specification rules to CLAUDE.md
+- add setup/onboarding specification (src/setup/README.md)
+- deduplicate tool code and remove dead stubs ([#98](https://github.com/nearai/ironclaw/pull/98))
+- Reformat architecture diagram in README ([#64](https://github.com/nearai/ironclaw/pull/64))
+- Add review discipline guidelines to CLAUDE.md ([#68](https://github.com/nearai/ironclaw/pull/68))
+- Bump MSRV to 1.92, add GCP deployment files ([#40](https://github.com/nearai/ironclaw/pull/40))
+- Add OpenAI-compatible HTTP API (/v1/chat/completions, /v1/models)   ([#31](https://github.com/nearai/ironclaw/pull/31))
+
+## [0.1.3](https://github.com/nearai/ironclaw/compare/v0.1.2...v0.1.3) - 2026-02-12
+
+### Other
+
+- Enabled builds caching during CI/CD
+- Disabled npm publishing as the name is already taken
+
 ## [0.1.2](https://github.com/nearai/ironclaw/compare/v0.1.1...v0.1.2) - 2026-02-12
 
 ### Other

CLAUDE.md

@@ -151,6 +151,12 @@ src/
 │       ├── rate_limiter.rs # Per-tool rate limiting
 │       └── storage.rs  # Linear memory persistence
 │
+├── db/                 # Database abstraction layer
+│   ├── mod.rs          # Database trait (~60 async methods)
+│   ├── postgres.rs     # PostgreSQL backend (delegates to Store + Repository)
+│   ├── libsql_backend.rs # libSQL/Turso backend (embedded SQLite)
+│   └── libsql_migrations.rs # SQLite-dialect schema (idempotent)
+│
 ├── workspace/          # Persistent memory system (OpenClaw-inspired)
 │   ├── mod.rs          # Workspace struct, memory operations
 │   ├── document.rs     # MemoryDocument, MemoryChunk, WorkspaceEntry
@@ -192,8 +198,9 @@ When designing new features or systems, always prefer generic/extensible archite
 
 ### Error Handling
 - Use `thiserror` for error types in `error.rs`
-- Never use `.unwrap()` in production code (tests are fine)
+- Never use `.unwrap()` or `.expect()` in production code (tests are fine)
 - Map errors with context: `.map_err(|e| SomeError::Variant { reason: e.to_string() })?`
+- Before committing, grep for `.unwrap()` and `.expect(` in changed files to catch violations mechanically
 
 ### Async
 - All I/O is async with tokio
@@ -201,6 +208,7 @@ When designing new features or systems, always prefer generic/extensible archite
 - Use `RwLock` for concurrent read/write access
 
 ### Traits for Extensibility
+- `Database` - Add new database backends (must implement all ~60 methods)
 - `Channel` - Add new input sources
 - `Tool` - Add new capabilities
 - `LlmProvider` - Add new LLM backends
@@ -248,7 +256,12 @@ Pending -> InProgress -> Completed -> Submitted -> Accepted
 
 Environment variables (see `.env.example`):
 ```bash
+# Database backend (default: postgres)
+DATABASE_BACKEND=postgres               # or "libsql" / "turso"
 DATABASE_URL=postgres://user:pass@localhost/ironclaw
+LIBSQL_PATH=~/.ironclaw/ironclaw.db    # libSQL local path (default)
+# LIBSQL_URL=libsql://xxx.turso.io    # Turso cloud (optional)
+# LIBSQL_AUTH_TOKEN=xxx                # Required with LIBSQL_URL
 
 # NEAR AI (required)
 NEARAI_SESSION_TOKEN=sess_...
@@ -308,7 +321,51 @@ Session tokens have the format `sess_xxx` (37 characters). They are authenticate
 
 ## Database
 
-Single migration in `migrations/V1__initial.sql`. Tables:
+IronClaw supports two database backends, selected at compile time via Cargo feature flags and at runtime via the `DATABASE_BACKEND` environment variable.
+
+**IMPORTANT: All new features that touch persistence MUST support both backends.** Implement the operation as a method on the `Database` trait in `src/db/mod.rs`, then add the implementation in both `src/db/postgres.rs` (delegate to Store/Repository) and `src/db/libsql_backend.rs` (native SQL).
+
+### Backends
+
+| Backend | Feature Flag | Default | Use Case |
+|---------|-------------|---------|----------|
+| PostgreSQL | `postgres` (default) | Yes | Production, existing deployments |
+| libSQL/Turso | `libsql` | No | Zero-dependency local mode, edge, Turso cloud |
+
+```bash
+# Build with PostgreSQL only (default)
+cargo build
+
+# Build with libSQL only
+cargo build --no-default-features --features libsql
+
+# Build with both backends available
+cargo build --features "postgres,libsql"
+```
+
+### Database Trait
+
+The `Database` trait (`src/db/mod.rs`) defines ~60 async methods covering all persistence:
+- Conversations, messages, metadata
+- Jobs, actions, LLM calls, estimation snapshots
+- Sandbox jobs, job events
+- Routines, routine runs
+- Tool failures, settings
+- Workspace: documents, chunks, hybrid search
+
+Both backends implement this trait. PostgreSQL delegates to the existing `Store` + `Repository`. libSQL implements native SQLite-dialect SQL.
+
+### Schema
+
+**PostgreSQL:** `migrations/V1__initial.sql` (351 lines). Uses pgvector for embeddings, tsvector for FTS, PL/pgSQL functions. Managed by `refinery`.
+
+**libSQL:** `src/db/libsql_migrations.rs` (consolidated schema, ~480 lines). Translates PG types:
+- `UUID` -> `TEXT`, `TIMESTAMPTZ` -> `TEXT` (ISO-8601), `JSONB` -> `TEXT`
+- `VECTOR(1536)` -> `F32_BLOB(1536)` with `libsql_vector_idx`
+- `tsvector`/`ts_rank_cd` -> FTS5 virtual table with sync triggers
+- PL/pgSQL functions -> SQLite triggers
+
+**Tables (both backends):**
 
 **Core:**
 - `conversations` - Multi-channel conversation tracking
@@ -320,12 +377,41 @@ Single migration in `migrations/V1__initial.sql`. Tables:
 
 **Workspace/Memory:**
 - `memory_documents` - Flexible path-based files (e.g., "context/vision.md", "daily/2024-01-15.md")
-- `memory_chunks` - Chunked content with FTS (tsvector) and vector (pgvector) indexes
+- `memory_chunks` - Chunked content with FTS and vector indexes
 - `heartbeat_state` - Periodic execution tracking
 
-Requires pgvector extension: `CREATE EXTENSION IF NOT EXISTS vector;`
+**Other:**
+- `routines`, `routine_runs` - Scheduled/reactive execution
+- `settings` - Per-user key-value settings
+- `tool_failures` - Self-repair tracking
+- `secrets`, `wasm_tools`, `tool_capabilities` - Extension infrastructure
 
-Run migrations: `refinery migrate -c refinery.toml`
+### Configuration
+
+```bash
+# Backend selection (default: postgres)
+DATABASE_BACKEND=libsql
+
+# PostgreSQL
+DATABASE_URL=postgres://user:pass@localhost/ironclaw
+
+# libSQL (embedded)
+LIBSQL_PATH=~/.ironclaw/ironclaw.db    # Default path
+
+# libSQL (Turso cloud sync)
+LIBSQL_URL=libsql://your-db.turso.io
+LIBSQL_AUTH_TOKEN=your-token            # Required when LIBSQL_URL is set
+```
+
+### Current Limitations (libSQL backend)
+
+- **Workspace/memory system** not yet wired through Database trait (requires Store migration)
+- **Secrets store** not yet available (still requires PostgresSecretsStore)
+- **Hybrid search** uses FTS5 only (vector search via libsql_vector_idx not yet implemented)
+- **Settings reload from DB** skipped (Config::from_db requires Store)
+- No incremental migration versioning (schema is CREATE IF NOT EXISTS, no ALTER TABLE support yet)
+- **No encryption at rest** -- The local SQLite database file stores conversation content, job data, workspace memory, and other application data in plaintext. Only secrets (API tokens, credentials) are encrypted via AES-256-GCM before storage. Users handling sensitive data should use full-disk encryption (FileVault, LUKS, BitLocker) or consider the PostgreSQL backend with TDE/encrypted storage.
+- **JSON merge patch vs path-targeted update** -- The libSQL backend uses RFC 7396 JSON Merge Patch (`json_patch`) for metadata updates, while PostgreSQL uses path-targeted `jsonb_set`. Merge patch replaces top-level keys entirely, which may drop nested keys not present in the patch. Callers should avoid relying on partial nested object updates in metadata fields.
 
 ## Safety Layer
 
@@ -387,6 +473,7 @@ Key test patterns:
 - ✅ **Claude Code mode** - Delegate jobs to Claude CLI inside containers
 - ✅ **Routines system** - Cron, event, webhook, and manual triggers with guardrails
 - ✅ **Extension management** - Install, auth, activate MCP/WASM extensions via CLI and web UI
+- ✅ **libSQL/Turso backend** - Database trait abstraction (`src/db/`), feature-gated dual backend support (postgres/libsql), embedded SQLite for zero-dependency local mode
 
 ## Adding a New Tool
 
@@ -543,6 +630,22 @@ RUST_LOG=ironclaw::agent=debug cargo run
 RUST_LOG=ironclaw=debug,tower_http=debug cargo run
 ```
 
+## Module Specifications
+
+Some modules have a `README.md` that serves as the authoritative specification
+for that module's behavior. When modifying code in a module that has a spec:
+
+1. **Read the spec first** before making changes
+2. **Code follows spec**: if the spec says X, the code must do X
+3. **Update both sides**: if you change behavior, update the spec to match;
+   if you're implementing a spec change, update the code to match
+4. **Spec is the tiebreaker**: when code and spec disagree, the spec is correct
+   (unless the spec is clearly outdated, in which case fix the spec first)
+
+| Module | Spec File |
+|--------|-----------|
+| `src/setup/` | `src/setup/README.md` |
+
 ## Code Style
 
 - Use `crate::` imports, not `super::`
@@ -551,6 +654,37 @@ RUST_LOG=ironclaw=debug,tower_http=debug cargo run
 - Keep functions focused, extract helpers when logic is reused
 - Comments for non-obvious logic only
 
+## Review & Fix Discipline
+
+Hard-won lessons from code review -- follow these when fixing bugs or addressing review feedback.
+
+### Fix the pattern, not just the instance
+When a reviewer flags a bug (e.g., TOCTOU race in INSERT + SELECT-back), search the entire codebase for all instances of that same pattern. A fix in `SecretsStore::create()` that doesn't also fix `WasmToolStore::store()` is half a fix.
+
+### Propagate architectural fixes to satellite types
+If a core type changes its concurrency model (e.g., `LibSqlBackend` switches to connection-per-operation), every type that was handed a resource from the old model (e.g., `LibSqlSecretsStore`, `LibSqlWasmToolStore` holding a single `Connection`) must also be updated. Grep for the old type across the codebase.
+
+### Schema translation is more than DDL
+When translating a database schema between backends (PostgreSQL to libSQL, etc.), check for:
+- **Indexes** -- diff `CREATE INDEX` statements between the two schemas
+- **Seed data** -- check for `INSERT INTO` in migrations (e.g., `leak_detection_patterns`)
+- **Semantic differences** -- document where SQL functions behave differently (e.g., `json_patch` vs `jsonb_set`)
+
+### Feature flag testing
+When adding feature-gated code, test compilation with each feature in isolation:
+```bash
+cargo check                                          # default features
+cargo check --no-default-features --features libsql  # libsql only
+cargo check --all-features                           # all features
+```
+Dead code behind the wrong `#[cfg]` gate will only show up when building with a single feature.
+
+### Mechanical verification before committing
+Run these checks on changed files before committing:
+- `grep -rnE '\.unwrap\(|\.expect\(' <files>` -- no panics in production
+- `grep -rn 'super::' <files>` -- use `crate::` imports
+- If you fixed a pattern bug, `grep` for other instances of that pattern across `src/`
+
 ## Workspace & Memory System
 
 Inspired by [OpenClaw](https://github.com/openclaw/openclaw), the workspace provides persistent memory for agents with a flexible filesystem-like structure.
@@ -625,7 +759,7 @@ Four tools for LLM use:
 
 ### Hybrid Search (RRF)
 
-Combines full-text search (PostgreSQL `ts_rank_cd`) and vector similarity (pgvector cosine) using Reciprocal Rank Fusion:
+Combines full-text search and vector similarity using Reciprocal Rank Fusion:
 
 ```
 score(d) = Σ 1/(k + rank(d)) for each method where d appears
@@ -633,6 +767,10 @@ score(d) = Σ 1/(k + rank(d)) for each method where d appears
 
 Default k=60. Results from both methods are combined, with documents appearing in both getting boosted scores.
 
+**Backend differences:**
+- **PostgreSQL:** `ts_rank_cd` for FTS, pgvector cosine distance for vectors, full RRF
+- **libSQL:** FTS5 for keyword search only (vector search via `libsql_vector_idx` not yet wired)
+
 ### Heartbeat System
 
 Proactive periodic execution (default: 30 minutes):

Cargo.toml

@@ -1,8 +1,16 @@
+[workspace]
+exclude = [
+    "channels-src/telegram",
+    "channels-src/slack",
+    "channels-src/whatsapp",
+    "tools-src/gmail",
+]
+
 [package]
 name = "ironclaw"
-version = "0.1.2"
+version = "0.5.0"
 edition = "2024"
-rust-version = "1.85"
+rust-version = "1.92"
 description = "Secure personal AI assistant that protects your data and expands its capabilities on the fly"
 authors = ["NEAR AI <[email protected]>"]
 license = "MIT OR Apache-2.0"
@@ -22,17 +30,20 @@ tokio-stream = { version = "0.1", features = ["sync"] }
 futures = "0.3"
 
 # HTTP client
-reqwest = { version = "0.12", default-features = false, features = ["json", "rustls-tls", "stream"] }
+reqwest = { version = "0.12", default-features = false, features = ["json", "rustls-tls-native-roots", "stream"] }
 
 # Serialization
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 
-# Database
-deadpool-postgres = "0.14"
-tokio-postgres = { version = "0.7", features = ["with-uuid-1", "with-chrono-0_4", "with-serde_json-1"] }
-postgres-types = { version = "0.2", features = ["with-serde_json-1"] }
-refinery = { version = "0.8", features = ["tokio-postgres"] }
+# Database - PostgreSQL (default, feature-gated)
+deadpool-postgres = { version = "0.14", optional = true }
+tokio-postgres = { version = "0.7", features = ["with-uuid-1", "with-chrono-0_4", "with-serde_json-1"], optional = true }
+postgres-types = { version = "0.2", features = ["with-serde_json-1"], optional = true }
+refinery = { version = "0.8", features = ["tokio-postgres"], optional = true }
+
+# Database - libSQL/Turso (optional embedded database)
+libsql = { version = "0.6", optional = true, default-features = false, features = ["core", "replication"] }
 
 # Error handling
 thiserror = "2"
@@ -48,7 +59,7 @@ dotenvy = "0.15"
 # Core types
 uuid = { version = "1", features = ["v4", "serde"] }
 chrono = { version = "0.4", features = ["serde"] }
-rust_decimal = { version = "1", features = ["serde", "serde-with-str", "db-tokio-postgres", "maths"] }
+rust_decimal = { version = "1", features = ["serde", "serde-with-str", "maths"] }
 rust_decimal_macros = "1"
 
 # Async traits
@@ -81,7 +92,8 @@ fs4 = "0.6"
 # Secrecy for sensitive values
 secrecy = { version = "0.10", features = ["serde"] }
 
-# URL encoding for OAuth flow
+# URL parsing and encoding
+url = "2"
 urlencoding = "2"
 
 # Open URLs in browser
@@ -89,7 +101,7 @@ open = "5"
 
 # Vector embeddings for semantic search
 # The postgres feature provides ToSql/FromSql for postgres-types (shared by tokio-postgres)
-pgvector = { version = "0.4", features = ["postgres"] }
+pgvector = { version = "0.4", features = ["postgres"], optional = true }
 
 # WASM sandbox for untrusted tool execution
 wasmtime = { version = "28", features = ["component-model"] }
@@ -102,6 +114,7 @@ hkdf = "0.12"
 sha2 = "0.10"
 blake3 = "1"
 rand = "0.8"
+subtle = "2"  # Constant-time comparisons for token validation
 
 # Multi-provider LLM support
 rig-core = "0.30"
@@ -134,9 +147,22 @@ pretty_assertions = "1"
 tempfile = "3"
 
 [features]
-default = []
+default = ["postgres", "libsql"]
+postgres = [
+    "dep:deadpool-postgres",
+    "dep:tokio-postgres",
+    "dep:postgres-types",
+    "dep:refinery",
+    "dep:pgvector",
+    "rust_decimal/db-tokio-postgres",
+]
+libsql = ["dep:libsql"]
 integration = []
 
+[[example]]
+name = "test_heartbeat"
+required-features = ["postgres"]
+
 # The profile that 'cargo dist' will build with
 [profile.dist]
 inherits = "release"
@@ -151,7 +177,7 @@ ci = "github"
 # The installers to generate for each app
 installers = ["shell", "powershell", "npm", "msi"]
 # Publish jobs to run in CI
-publish-jobs = ["npm"]
+publish-jobs = []
 # Target platforms to build apps for (Rust target-triple syntax)
 targets = [
     "aarch64-apple-darwin",
@@ -165,11 +191,13 @@ windows-archive = ".tar.gz"
 # The archive format to use for non-windows builds (defaults .tar.xz)
 unix-archive = ".tar.gz"
 # Which actions to run on pull requests
-pr-run-mode = "upload"
+pr-run-mode = "skip"
 # Path that installers should place binaries in
 install-path = "CARGO_HOME"
 # Whether to install an updater program
 install-updater = true
+# Cache intermediate build artifacts to speed up the release pipelines
+cache-builds = true
 
 [workspace.metadata.dist.github-custom-runners]
 aarch64-unknown-linux-gnu = "ubuntu-24.04-arm"

Dockerfile

@@ -0,0 +1,46 @@
+# Multi-stage Dockerfile for the IronClaw agent (cloud deployment).
+#
+# Build:
+#   docker build --platform linux/amd64 -t ironclaw:latest .
+#
+# Run:
+#   docker run --env-file .env -p 3000:3000 ironclaw:latest
+
+# Stage 1: Build
+FROM rust:1.92-slim-bookworm AS builder
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    pkg-config libssl-dev cmake gcc g++ \
+    && rm -rf /var/lib/apt/lists/*
+
+WORKDIR /app
+
+# Copy manifests first for layer caching
+COPY Cargo.toml Cargo.lock ./
+
+# Copy source and build artifacts
+COPY src/ src/
+COPY migrations/ migrations/
+COPY wit/ wit/
+
+RUN cargo build --release --bin ironclaw
+
+# Stage 2: Runtime
+FROM debian:bookworm-slim
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    ca-certificates libssl3 \
+    && rm -rf /var/lib/apt/lists/*
+
+COPY --from=builder /app/target/release/ironclaw /usr/local/bin/ironclaw
+COPY --from=builder /app/migrations /app/migrations
+
+# Non-root user
+RUN useradd -m -u 1000 -s /bin/bash ironclaw
+USER ironclaw
+
+EXPOSE 3000
+
+ENV RUST_LOG=ironclaw=info
+
+ENTRYPOINT ["ironclaw"]

Dockerfile.worker

@@ -9,7 +9,7 @@
 # The image includes common development tools so workers can build software,
 # run tests, and execute shell commands.
 
-FROM rust:1.85-bookworm AS builder
+FROM rust:1.92-bookworm AS builder
 
 WORKDIR /build
 COPY . .
@@ -40,7 +40,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 ENV RUSTUP_HOME=/usr/local/rustup \
     CARGO_HOME=/usr/local/cargo \
     PATH=/usr/local/cargo/bin:$PATH
-RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y --default-toolchain 1.85.0 \
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y --default-toolchain 1.92.0 \
     && chmod -R a+r /usr/local/rustup /usr/local/cargo
 
 # Install Claude Code CLI (for claude-bridge mode)

FEATURE_PARITY.md

@@ -37,7 +37,7 @@ This document tracks feature parity between IronClaw (Rust implementation) and O
 | Session management/routing | ✅ | ✅ | SessionManager exists |
 | Configuration hot-reload | ✅ | ❌ | |
 | Network modes (loopback/LAN/remote) | ✅ | 🚧 | HTTP only |
-| OpenAI-compatible HTTP API | ✅ | ❌ | /v1/chat/completions |
+| OpenAI-compatible HTTP API | ✅ | ✅ | /v1/chat/completions |
 | Canvas hosting | ✅ | ❌ | Agent-driven UI |
 | Gateway lock (PID-based) | ✅ | ❌ | |
 | launchd/systemd integration | ✅ | ❌ | |
@@ -112,7 +112,7 @@ This document tracks feature parity between IronClaw (Rust implementation) and O
 | `pairing` | ✅ | ✅ | - | list/approve for channel DM pairing |
 | `nodes` | ✅ | ❌ | P3 | Device management |
 | `plugins` | ✅ | ❌ | P3 | Plugin management |
-| `hooks` | ✅ | ❌ | P2 | Lifecycle hooks |
+| `hooks` | ✅ | ✅ | P2 | Lifecycle hooks |
 | `cron` | ✅ | ❌ | P2 | Scheduled jobs |
 | `webhooks` | ✅ | ❌ | P3 | Webhook config |
 | `message send` | ✅ | ❌ | P2 | Send to channels |
@@ -133,7 +133,7 @@ This document tracks feature parity between IronClaw (Rust implementation) and O
 |---------|----------|----------|-------|
 | Pi agent runtime | ✅ | ➖ | IronClaw uses custom runtime |
 | RPC-based execution | ✅ | ✅ | Orchestrator/worker pattern |
-| Multi-provider failover | ✅ | ❌ | Provider fallback chains |
+| Multi-provider failover | ✅ | ✅ | `FailoverProvider` tries providers sequentially on retryable errors |
 | Per-sender sessions | ✅ | ✅ | |
 | Global sessions | ✅ | ❌ | Optional shared context |
 | Session pruning | ✅ | ❌ | Auto cleanup old sessions |
@@ -164,7 +164,7 @@ This document tracks feature parity between IronClaw (Rust implementation) and O
 | AWS Bedrock | ✅ | ❌ | P3 | |
 | Google Gemini | ✅ | ❌ | P3 | |
 | OpenRouter | ✅ | ❌ | P3 | |
-| Ollama (local) | ✅ | ❌ | P2 | Local models |
+| Ollama (local) | ✅ | ✅ | - | via `rig::providers::ollama` (full support) |
 | node-llama-cpp | ✅ | ➖ | - | N/A for Rust |
 | llama.cpp (native) | ❌ | 🔮 | P3 | Rust bindings |
 
@@ -173,8 +173,8 @@ This document tracks feature parity between IronClaw (Rust implementation) and O
 | Feature | OpenClaw | IronClaw | Notes |
 |---------|----------|----------|-------|
 | Auto-discovery | ✅ | ❌ | |
-| Failover chains | ✅ | ❌ | Provider fallback |
-| Cooldown management | ✅ | ❌ | Skip failed providers |
+| Failover chains | ✅ | ✅ | `FailoverProvider` with configurable `fallback_model` |
+| Cooldown management | ✅ | ✅ | Lock-free per-provider cooldown in `FailoverProvider` |
 | Per-session model override | ✅ | ✅ | Model selector in TUI |
 | Model selection UI | ✅ | ✅ | TUI keyboard shortcut |
 
@@ -323,14 +323,14 @@ This document tracks feature parity between IronClaw (Rust implementation) and O
 | Cron jobs | ✅ | ✅ | - | Routines with cron trigger |
 | Timezone support | ✅ | ✅ | - | Via cron expressions |
 | One-shot/recurring jobs | ✅ | ✅ | - | Manual + cron triggers |
-| `beforeInbound` hook | ✅ | ❌ | P2 | |
-| `beforeOutbound` hook | ✅ | ❌ | P2 | |
-| `beforeToolCall` hook | ✅ | ❌ | P2 | |
+| `beforeInbound` hook | ✅ | ✅ | P2 | |
+| `beforeOutbound` hook | ✅ | ✅ | P2 | |
+| `beforeToolCall` hook | ✅ | ✅ | P2 | |
 | `onMessage` hook | ✅ | ✅ | - | Routines with event trigger |
-| `onSessionStart` hook | ✅ | ❌ | P2 | |
-| `onSessionEnd` hook | ✅ | ❌ | P2 | |
+| `onSessionStart` hook | ✅ | ✅ | P2 | |
+| `onSessionEnd` hook | ✅ | ✅ | P2 | |
 | `transcribeAudio` hook | ✅ | ❌ | P3 | |
-| `transformResponse` hook | ✅ | ❌ | P2 | |
+| `transformResponse` hook | ✅ | ✅ | P2 | |
 | Bundled hooks | ✅ | ❌ | P2 | |
 | Plugin hooks | ✅ | ❌ | P3 | |
 | Workspace hooks | ✅ | ❌ | P2 | Inline code |
@@ -419,15 +419,11 @@ This document tracks feature parity between IronClaw (Rust implementation) and O
 - ❌ Slack channel (real implementation)
 - ✅ Telegram channel (WASM, DM pairing, caption, /start)
 - ❌ WhatsApp channel
-- ❌ Multi-provider failover
-- ❌ Hooks system (beforeInbound, beforeToolCall, etc.)
+- ✅ Multi-provider failover (`FailoverProvider` with retryable error classification)
+- ✅ Hooks system (beforeInbound, beforeToolCall, beforeOutbound, onSessionStart, onSessionEnd, transformResponse)
 
 ### P2 - Medium Priority
-- ❌ Cron job scheduling
-- ❌ Web Control UI
-- ❌ WebChat channel
-- 🚧 Media handling (caption support; no image/PDF processing)
-- ❌ CLI subcommands (config, status, memory, doctor)
+- ❌ Media handling (images, PDFs)
 - ❌ Ollama/local model support
 - ❌ Configuration hot-reload
 - ❌ Webhook trigger endpoint in web gateway

README.md

@@ -99,22 +99,6 @@ curl --proto '=https' --tlsv1.2 -LsSf https://github.com/nearai/ironclaw/release
 ```
 </details>
 
-<details>
-  <summary>Run via npx (Node.js on Windows, Linux, macOS)</summary>
-
-```sh
-npx ironclaw
-```
-</details>
-
-<details>
-  <summary>Use in package.json scripts (Node.js on Windows, Linux, macOS)</summary>
-
-```sh
-npm install ironclaw
-```
-</details>
-
 <details>
   <summary>Compile the source code (Cargo on Windows, Linux, macOS)</summary>
 
@@ -197,42 +181,42 @@ External content passes through multiple security layers:
 ## Architecture
 
 ```
-┌────────────────────────────────────────────────────────────────────┐
-│                          Channels                                  │
-│  ┌──────┐  ┌──────┐  ┌─────────────┐  ┌─────────────┐            │
-│  │ REPL │  │ HTTP │  │WASM Channels│  │ Web Gateway │            │
-│  └──┬───┘  └──┬───┘  └──────┬──────┘  │ (SSE + WS) │            │
-│     │         │              │         └──────┬──────┘            │
-│     └─────────┴──────────────┴────────────────┘                   │
-│                              │                                     │
-│                    ┌─────────▼─────────┐                          │
-│                    │    Agent Loop     │  Intent routing           │
-│                    └────┬─────────┬────┘                          │
-│                         │         │                                │
-│              ┌──────────▼───┐  ┌──▼──────────────┐               │
-│              │  Scheduler   │  │ Routines Engine  │               │
-│              │(parallel jobs)│  │(cron, event, wh) │               │
-│              └──────┬───────┘  └────────┬─────────┘               │
-│                     │                   │                          │
-│       ┌─────────────┼───────────────────┘                         │
-│       │             │                                              │
-│   ┌───▼────┐   ┌────▼────────────────┐                           │
-│   │ Local  │   │    Orchestrator     │                           │
-│   │Workers │   │  ┌───────────────┐  │                           │
-│   │(in-proc)│   │  │ Docker Sandbox│  │                           │
-│   └───┬────┘   │  │   Containers  │  │                           │
-│       │        │  │ ┌───────────┐ │  │                           │
-│       │        │  │ │Worker / CC│ │  │                           │
-│       │        │  │ └───────────┘ │  │                           │
-│       │        │  └───────────────┘  │                           │
-│       │        └─────────┬───────────┘                           │
-│       └──────────────────┤                                        │
-│                          │                                        │
-│              ┌───────────▼──────────┐                             │
-│              │    Tool Registry     │                             │
-│              │  Built-in, MCP, WASM │                             │
-│              └──────────────────────┘                             │
-└────────────────────────────────────────────────────────────────────┘
+┌────────────────────────────────────────────────────────────────┐
+│                          Channels                              │
+│  ┌──────┐  ┌──────┐   ┌─────────────┐  ┌─────────────┐         │
+│  │ REPL │  │ HTTP │   │WASM Channels│  │ Web Gateway │         │
+│  └──┬───┘  └──┬───┘   └──────┬──────┘  │ (SSE + WS)  │         │
+│     │         │              │         └──────┬──────┘         │
+│     └─────────┴──────────────┴────────────────┘                │
+│                              │                                 │
+│                    ┌─────────▼─────────┐                       │
+│                    │    Agent Loop     │  Intent routing       │
+│                    └────┬──────────┬───┘                       │
+│                         │          │                           │
+│              ┌──────────▼────┐  ┌──▼───────────────┐           │
+│              │  Scheduler    │  │ Routines Engine  │           │
+│              │(parallel jobs)│  │(cron, event, wh) │           │
+│              └──────┬────────┘  └────────┬─────────┘           │
+│                     │                    │                     │
+│       ┌─────────────┼────────────────────┘                     │
+│       │             │                                          │
+│   ┌───▼─────┐  ┌────▼────────────────┐                         │
+│   │ Local   │  │    Orchestrator     │                         │
+│   │Workers  │  │  ┌───────────────┐  │                         │
+│   │(in-proc)│  │  │ Docker Sandbox│  │                         │
+│   └───┬─────┘  │  │   Containers  │  │                         │
+│       │        │  │ ┌───────────┐ │  │                         │
+│       │        │  │ │Worker / CC│ │  │                         │
+│       │        │  │ └───────────┘ │  │                         │
+│       │        │  └───────────────┘  │                         │
+│       │        └─────────┬───────────┘                         │
+│       └──────────────────┤                                     │
+│                          │                                     │
+│              ┌───────────▼──────────┐                          │
+│              │    Tool Registry     │                          │
+│              │  Built-in, MCP, WASM │                          │
+│              └──────────────────────┘                          │
+└────────────────────────────────────────────────────────────────┘
 ```
 
 ### Core Components

channels-src/discord/Cargo.toml

@@ -0,0 +1,23 @@
+[package]
+name = "discord-channel"
+version = "0.1.0"
+edition = "2021"
+description = "Discord channel for IronClaw"
+license = "MIT OR Apache-2.0"
+publish = false
+
+[dependencies]
+serde = { version = "1.0", features = ["derive"] }
+serde_json = "1.0"
+wit-bindgen = "0.41.0"
+
+[lib]
+crate-type = ["cdylib"]
+
+[profile.release]
+strip = true
+opt-level = "s"
+lto = true
+codegen-units = 1
+
+

channels-src/discord/README.md

@@ -0,0 +1,121 @@
+# Discord Channel for IronClaw
+
+WASM channel for Discord integration - handle slash commands and button interactions via webhooks.
+
+## Features
+
+- **Slash Commands** - Process Discord slash commands
+- **Button Interactions** - Handle button clicks
+- **Thread Support** - Respond in threads
+- **DM Support** - Handle direct messages
+
+## Setup
+
+1. Create a Discord Application at <https://discord.com/developers/applications>
+2. Create a Bot and get the token
+3. Set up Interactions URL to point to your IronClaw instance
+4. Copy the Application ID and Public Key
+5. Store in IronClaw secrets:
+
+   ```bash
+   ironclaw secret set discord_bot_token YOUR_BOT_TOKEN
+   ```
+
+   **Note:** The `discord_bot_token` secret is the only value read directly by this
+   Discord channel WASM component. The `discord_app_id` and `discord_public_key`
+   secrets are used by the IronClaw host (for example, to verify Discord
+   interaction signatures and manage slash command registration) and are not
+   accessed from the WASM module itself.
+
+## Discord Configuration
+
+### Register Slash Commands
+
+```bash
+curl -X POST \
+  -H "Authorization: Bot YOUR_BOT_TOKEN" \
+  -H "Content-Type: application/json" \
+  https://discord.com/api/v10/applications/YOUR_APP_ID/commands \
+  -d '{
+    "name": "ask",
+    "description": "Ask the AI agent",
+    "options": [{
+      "name": "question",
+      "description": "Your question",
+      "type": 3,
+      "required": true
+    }]
+  }'
+```
+
+### Set Interactions Endpoint
+
+In your Discord app settings, set:
+
+- Interactions Endpoint URL: `https://your-ironclaw.com/webhook/discord`
+
+## Usage Examples
+
+### Slash Command
+
+User types: `/ask question: What is the weather?`
+
+The agent receives:
+
+```text
+User: @username
+Content: /ask question: What is the weather?
+```
+
+### Button Click
+
+When a user clicks a button in a message, the agent receives:
+
+```text
+User: @username
+Content: [Button clicked] Original message content
+```
+
+## Error Handling
+
+If an internal error occurs (e.g., metadata serialization failure), the tool attempts to send an ephemeral message to the user:
+
+```text
+❌ Internal Error: Failed to process command metadata.
+```
+
+Check the host logs for detailed error information.
+
+## Advanced Usage
+
+### Embeds
+
+To send embeds, include an `embeds` array in the `metadata_json` field of the agent's response. The structure should match the Discord API `embed` object.
+
+## Troubleshooting
+
+### "Invalid Signature"
+
+- Check that `discord_public_key` is set correctly in IronClaw secrets.
+- This validation happens on the host before reaching the WASM.
+
+### "401 Unauthorized"
+
+- Check that `discord_bot_token` is set correctly in IronClaw secrets.
+- Ensure the bot is added to the server.
+
+### "Interaction Failed"
+
+- The interaction might have timed out (Discord requires a response within 3 seconds).
+- The `interactions_endpoint_url` might be unreachable.
+
+## Building
+
+```bash
+cd channels-src/discord
+cargo build --target wasm32-wasi --release
+```
+
+## License
+
+MIT/Apache-2.0

channels-src/discord/discord.capabilities.json

@@ -0,0 +1,39 @@
+{
+  "type": "channel",
+  "name": "discord",
+  "description": "Discord Gateway/Webhook channel for handling slash commands, buttons, and messages",
+  "capabilities": {
+    "http": {
+      "allowlist": [
+        { "host": "discord.com", "path_prefix": "/api/v10" }
+      ],
+      "credentials": {
+        "discord_bot_token": {
+          "secret_name": "discord_bot_token",
+          "location": { "type": "header", "header_name": "Authorization", "prefix": "Bot " },
+          "host_patterns": ["discord.com"]
+        }
+      },
+      "rate_limit": {
+        "requests_per_minute": 60,
+        "requests_per_hour": 3600
+      }
+    },
+    "secrets": {
+      "allowed_names": ["discord_bot_token", "discord_*"]
+    },
+    "channel": {
+      "allowed_paths": ["/webhook/discord"],
+      "allow_polling": false,
+      "callback_timeout_secs": 45,
+      "workspace_prefix": "channels/discord/",
+      "emit_rate_limit": {
+        "messages_per_minute": 100,
+        "messages_per_hour": 5000
+      }
+    }
+  },
+  "config": {
+    "require_signature_verification": true
+  }
+}

channels-src/discord/src/lib.rs

@@ -0,0 +1,476 @@
+//! Discord Gateway/Webhook channel for IronClaw.
+//!
+//! This WASM component implements the channel interface for handling Discord
+//! interactions via webhooks and sending messages back to Discord.
+//!
+//! # Features
+//!
+//! - URL verification for Discord interactions
+//! - Slash command handling
+//! - Message event parsing (@mentions, DMs)
+//! - Thread support for conversations
+//! - Response posting via Discord Web API
+//! - Automatic message truncation (> 2000 chars)
+//!
+//! # Security
+//!
+//! - Signature validation is handled by the host (webhook secrets)
+//! - Bot token is injected by host during HTTP requests
+//! - WASM never sees raw credentials
+
+wit_bindgen::generate!({
+    world: "sandboxed-channel",
+    path: "../../wit/channel.wit",
+});
+
+use serde::{Deserialize, Serialize};
+
+use exports::near::agent::channel::{
+    AgentResponse, ChannelConfig, Guest, HttpEndpointConfig, IncomingHttpRequest,
+    OutgoingHttpResponse, StatusUpdate,
+};
+use near::agent::channel_host::{self, EmittedMessage};
+
+/// Discord interaction wrapper.
+#[derive(Debug, Deserialize)]
+struct DiscordInteraction {
+    /// Interaction type (1=Ping, 2=ApplicationCommand, 3=MessageComponent)
+    #[serde(rename = "type")]
+    interaction_type: u8,
+
+    /// Interaction ID
+    id: String,
+
+    /// Application ID
+    application_id: String,
+
+    /// Guild ID (if in server)
+    #[allow(dead_code)] // Part of API payload, currently unused
+    guild_id: Option<String>,
+
+    /// Channel ID
+    channel_id: Option<String>,
+
+    /// Member info (if in server)
+    member: Option<DiscordMember>,
+
+    /// User info (if DM)
+    user: Option<DiscordUser>,
+
+    /// Command data (for slash commands)
+    data: Option<DiscordCommandData>,
+
+    /// Message (for component interactions)
+    message: Option<DiscordMessage>,
+
+    /// Token for responding
+    token: String,
+}
+
+#[derive(Debug, Deserialize, Clone)]
+struct DiscordMember {
+    user: DiscordUser,
+    #[allow(dead_code)] // Part of API payload, currently unused
+    nick: Option<String>,
+}
+
+#[derive(Debug, Deserialize, Clone)]
+struct DiscordUser {
+    id: String,
+    username: String,
+    global_name: Option<String>,
+}
+
+#[derive(Debug, Deserialize, Clone)]
+struct DiscordCommandData {
+    #[allow(dead_code)] // Part of API payload, currently unused
+    id: String,
+    name: String,
+    options: Option<Vec<DiscordCommandOption>>,
+}
+
+#[derive(Debug, Deserialize, Clone)]
+struct DiscordCommandOption {
+    name: String,
+    value: serde_json::Value,
+}
+
+#[derive(Debug, Deserialize, Clone)]
+struct DiscordMessage {
+    #[allow(dead_code)] // Part of API payload, currently unused
+    id: String,
+    content: String,
+    channel_id: String,
+    #[allow(dead_code)] // Part of API payload, currently unused
+    author: DiscordUser,
+}
+
+/// Metadata stored with emitted messages for response routing.
+#[derive(Debug, Serialize, Deserialize)]
+struct DiscordMessageMetadata {
+    /// Discord channel ID
+    channel_id: String,
+
+    /// Interaction ID for followups
+    interaction_id: String,
+
+    /// Interaction token for responding
+    token: String,
+
+    /// Application ID
+    application_id: String,
+
+    /// Thread ID (for forum threads)
+    thread_id: Option<String>,
+}
+
+struct DiscordChannel;
+
+impl Guest for DiscordChannel {
+    fn on_start(_config_json: String) -> Result<ChannelConfig, String> {
+        channel_host::log(channel_host::LogLevel::Info, "Discord channel starting");
+
+        Ok(ChannelConfig {
+            display_name: "Discord".to_string(),
+            http_endpoints: vec![HttpEndpointConfig {
+                path: "/webhook/discord".to_string(),
+                methods: vec!["POST".to_string()],
+                require_secret: true,
+            }],
+            poll: None,
+        })
+    }
+
+    fn on_http_request(req: IncomingHttpRequest) -> OutgoingHttpResponse {
+        let body_str = match std::str::from_utf8(&req.body) {
+            Ok(s) => s,
+            Err(_) => {
+                return json_response(400, serde_json::json!({"error": "Invalid UTF-8 body"}));
+            }
+        };
+
+        let interaction: DiscordInteraction = match serde_json::from_str(body_str) {
+            Ok(i) => i,
+            Err(e) => {
+                channel_host::log(
+                    channel_host::LogLevel::Error,
+                    &format!("Failed to parse Discord interaction: {}", e),
+                );
+                return json_response(400, serde_json::json!({"error": "Invalid interaction"}));
+            }
+        };
+
+        match interaction.interaction_type {
+            // Ping - Discord verification
+            1 => {
+                channel_host::log(channel_host::LogLevel::Info, "Responding to Discord ping");
+                json_response(200, serde_json::json!({"type": 1}))
+            }
+
+            // Application Command (slash command)
+            2 => {
+                handle_slash_command(&interaction);
+                json_response(
+                    200,
+                    serde_json::json!({
+                        "type": 5,
+                        "data": {
+                            "content": "🤔 Thinking..."
+                        }
+                    }),
+                )
+            }
+
+            // Message Component (buttons, selects)
+            3 => {
+                if let Some(ref message) = interaction.message {
+                    handle_message_component(&interaction, message);
+                }
+                json_response(200, serde_json::json!({"type": 6}))
+            }
+
+            _ => {
+                channel_host::log(
+                    channel_host::LogLevel::Warn,
+                    &format!(
+                        "Unknown Discord interaction type: {}",
+                        interaction.interaction_type
+                    ),
+                );
+                json_response(200, serde_json::json!({"type": 6}))
+            }
+        }
+    }
+
+    fn on_poll() {}
+
+    fn on_respond(response: AgentResponse) -> Result<(), String> {
+        let metadata: DiscordMessageMetadata = serde_json::from_str(&response.metadata_json)
+            .map_err(|e| format!("Failed to parse metadata: {}", e))?;
+
+        // Use webhook endpoint for followup
+        let url = format!(
+            "https://discord.com/api/v10/webhooks/{}/{}",
+            metadata.application_id, metadata.token
+        );
+
+        // Truncate content to 2000 characters to comply with Discord limits
+        let content = truncate_message(&response.content);
+
+        let mut payload = serde_json::json!({
+            "content": content,
+        });
+
+        // Check for embeds in metadata
+        if let Ok(meta_json) = serde_json::from_str::<serde_json::Value>(&response.metadata_json) {
+            if let Some(embeds) = meta_json.get("embeds") {
+                payload["embeds"] = embeds.clone();
+            }
+        }
+
+        let payload_bytes =
+            serde_json::to_vec(&payload).map_err(|e| format!("Failed to serialize: {}", e))?;
+
+        let headers = serde_json::json!({
+            "Content-Type": "application/json"
+        });
+
+        let result = channel_host::http_request(
+            "POST",
+            &url,
+            &headers.to_string(),
+            Some(&payload_bytes),
+            None,
+        );
+
+        match result {
+            Ok(http_response) => {
+                if http_response.status >= 200 && http_response.status < 300 {
+                    channel_host::log(channel_host::LogLevel::Debug, "Posted followup to Discord");
+                    Ok(())
+                } else {
+                    let body_str = String::from_utf8_lossy(&http_response.body);
+                    Err(format!(
+                        "Discord API error: {} - {}",
+                        http_response.status, body_str
+                    ))
+                }
+            }
+            Err(e) => Err(format!("HTTP request failed: {}", e)),
+        }
+    }
+
+    fn on_status(_update: StatusUpdate) {}
+
+    fn on_shutdown() {
+        channel_host::log(
+            channel_host::LogLevel::Info,
+            "Discord channel shutting down",
+        );
+    }
+}
+
+fn handle_slash_command(interaction: &DiscordInteraction) {
+    let user = interaction
+        .member
+        .as_ref()
+        .map(|m| &m.user)
+        .or(interaction.user.as_ref());
+    let user_id = user.map(|u| u.id.clone()).unwrap_or_default();
+    let user_name = user
+        .map(|u| {
+            u.global_name
+                .as_ref()
+                .filter(|s| !s.is_empty())
+                .unwrap_or(&u.username)
+                .clone()
+        })
+        .unwrap_or_default();
+
+    let channel_id = interaction.channel_id.clone().unwrap_or_default();
+
+    let command_name = interaction
+        .data
+        .as_ref()
+        .map(|d| d.name.clone())
+        .unwrap_or_default();
+    let options = interaction.data.as_ref().and_then(|d| d.options.clone());
+
+    let content = if let Some(opts) = options {
+        let opt_str = opts
+            .iter()
+            .map(|o| format!("{}: {}", o.name, o.value))
+            .collect::<Vec<_>>()
+            .join(", ");
+        format!("/{} {}", command_name, opt_str)
+    } else {
+        format!("/{}", command_name)
+    };
+
+    let metadata = DiscordMessageMetadata {
+        channel_id: channel_id.clone(),
+        interaction_id: interaction.id.clone(),
+        token: interaction.token.clone(),
+        application_id: interaction.application_id.clone(),
+        thread_id: None,
+    };
+
+    let metadata_json = match serde_json::to_string(&metadata) {
+        Ok(json) => json,
+        Err(e) => {
+            channel_host::log(
+                channel_host::LogLevel::Error,
+                &format!("Failed to serialize metadata: {}", e),
+            );
+            // Attempt to notify user of internal error
+            let url = format!(
+                "https://discord.com/api/v10/webhooks/{}/{}",
+                interaction.application_id, interaction.token
+            );
+            let payload = serde_json::json!({
+                "content": "❌ Internal Error: Failed to process command metadata.",
+                "flags": 64 // Ephemeral
+            });
+            let _ = channel_host::http_request(
+                "POST",
+                &url,
+                &serde_json::json!({"Content-Type": "application/json"}).to_string(),
+                Some(&serde_json::to_vec(&payload).unwrap_or_default()),
+                None,
+            );
+            return;
+        }
+    };
+
+    channel_host::emit_message(&EmittedMessage {
+        user_id,
+        user_name: Some(user_name),
+        content,
+        thread_id: None,
+        metadata_json,
+    });
+}
+
+fn handle_message_component(interaction: &DiscordInteraction, message: &DiscordMessage) {
+    // Check member first (for server contexts), then user (for DMs)
+    let user = interaction
+        .member
+        .as_ref()
+        .map(|m| &m.user)
+        .or(interaction.user.as_ref());
+    let user_id = user.map(|u| u.id.clone()).unwrap_or_default();
+    let user_name = user
+        .map(|u| {
+            u.global_name
+                .as_ref()
+                .filter(|s| !s.is_empty())
+                .unwrap_or(&u.username)
+                .clone()
+        })
+        .unwrap_or_default();
+
+    let channel_id = message.channel_id.clone();
+
+    let metadata = DiscordMessageMetadata {
+        channel_id: channel_id.clone(),
+        interaction_id: interaction.id.clone(),
+        token: interaction.token.clone(),
+        application_id: interaction.application_id.clone(),
+        thread_id: None,
+    };
+
+    let metadata_json = match serde_json::to_string(&metadata) {
+        Ok(json) => json,
+        Err(e) => {
+            channel_host::log(
+                channel_host::LogLevel::Error,
+                &format!("Failed to serialize metadata: {}", e),
+            );
+            return; // Don't emit message if metadata can't be serialized
+        }
+    };
+
+    channel_host::emit_message(&EmittedMessage {
+        user_id,
+        user_name: Some(user_name),
+        content: format!("[Button clicked] {}", message.content),
+        thread_id: None,
+        metadata_json,
+    });
+}
+
+fn json_response(status: u16, value: serde_json::Value) -> OutgoingHttpResponse {
+    let body = serde_json::to_vec(&value).unwrap_or_default();
+    let headers = serde_json::json!({"Content-Type": "application/json"});
+
+    OutgoingHttpResponse {
+        status,
+        headers_json: headers.to_string(),
+        body,
+    }
+}
+
+export!(DiscordChannel);
+
+fn truncate_message(content: &str) -> String {
+    if content.len() <= 2000 {
+        content.to_string()
+    } else {
+        let max_bytes = 1990;
+        let cutoff = content
+            .char_indices()
+            .map(|(i, c)| i + c.len_utf8())
+            .take_while(|&end| end <= max_bytes)
+            .last()
+            .unwrap_or(0);
+        let mut truncated = content[..cutoff].to_string();
+        truncated.push_str("\n... (truncated)");
+        truncated
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_truncate_message() {
+        let short = "Hello world";
+        assert_eq!(truncate_message(short), short);
+
+        let long = "a".repeat(2005);
+        let truncated = truncate_message(&long);
+        assert_eq!(truncated.len(), 2006); // 1990 + 16 chars suffix
+        assert!(truncated.ends_with("\n... (truncated)"));
+
+        // Test with multibyte characters (Euro sign is 3 bytes)
+        // 1000 chars * 3 bytes = 3000 bytes
+        let multi = "€".repeat(1000);
+        let truncated_multi = truncate_message(&multi);
+
+        // 1990 bytes limit. 1990 / 3 = 663 with remainder 1.
+        // Should truncate at 663 chars (1989 bytes).
+        // Suffix is 16 bytes. Total: 1989 + 16 = 2005 bytes.
+        assert!(truncated_multi.len() <= 2006);
+        assert!(truncated_multi.len() >= 2006 - 4); // Allow for max utf8 char width variance
+        assert!(truncated_multi.ends_with("\n... (truncated)"));
+
+        let content_part = &truncated_multi[..truncated_multi.len() - 16];
+        assert!(content_part.chars().all(|c| c == '€'));
+    }
+
+    #[test]
+    fn test_metadata_serialization() {
+        let metadata = DiscordMessageMetadata {
+            channel_id: "123".into(),
+            interaction_id: "456".into(),
+            token: "abc".into(),
+            application_id: "789".into(),
+            thread_id: None,
+        };
+        let json = serde_json::to_string(&metadata).unwrap();
+        let parsed: DiscordMessageMetadata = serde_json::from_str(&json).unwrap();
+        assert_eq!(parsed.channel_id, "123");
+        assert_eq!(parsed.interaction_id, "456");
+    }
+}

channels-src/slack/src/lib.rs

@@ -338,7 +338,13 @@ fn emit_message(
         team_id,
     };
 
-    let metadata_json = serde_json::to_string(&metadata).unwrap_or_else(|_| "{}".to_string());
+    let metadata_json = serde_json::to_string(&metadata).unwrap_or_else(|e| {
+        channel_host::log(
+            channel_host::LogLevel::Error,
+            &format!("Failed to serialize Slack metadata: {}", e),
+        );
+        "{}".to_string()
+    });
 
     // Strip @ mentions of the bot from the text for cleaner messages
     let cleaned_text = strip_bot_mention(&text);
@@ -366,7 +372,13 @@ fn strip_bot_mention(text: &str) -> String {
 
 /// Create a JSON HTTP response.
 fn json_response(status: u16, value: serde_json::Value) -> OutgoingHttpResponse {
-    let body = serde_json::to_vec(&value).unwrap_or_default();
+    let body = serde_json::to_vec(&value).unwrap_or_else(|e| {
+        channel_host::log(
+            channel_host::LogLevel::Error,
+            &format!("Failed to serialize JSON response: {}", e),
+        );
+        Vec::new()
+    });
     let headers = serde_json::json!({"Content-Type": "application/json"});
 
     OutgoingHttpResponse {

channels-src/telegram/src/lib.rs

@@ -285,11 +285,7 @@ impl Guest for TelegramChannel {
         }
 
         // Persist dm_policy and allow_from for DM pairing in handle_message
-        let dm_policy = config
-            .dm_policy
-            .as_deref()
-            .unwrap_or("pairing")
-            .to_string();
+        let dm_policy = config.dm_policy.as_deref().unwrap_or("pairing").to_string();
         let _ = channel_host::workspace_write(DM_POLICY_PATH, &dm_policy);
 
         let allow_from_json = serde_json::to_string(&config.allow_from.unwrap_or_default())
@@ -844,8 +840,8 @@ fn send_pairing_reply(chat_id: i64, code: &str) -> Result<(), String> {
         "parse_mode": "Markdown",
     });
 
-    let payload_bytes = serde_json::to_vec(&payload)
-        .map_err(|e| format!("Failed to serialize payload: {}", e))?;
+    let payload_bytes =
+        serde_json::to_vec(&payload).map_err(|e| format!("Failed to serialize payload: {}", e))?;
 
     let headers = serde_json::json!({
         "Content-Type": "application/json"
@@ -856,6 +852,7 @@ fn send_pairing_reply(chat_id: i64, code: &str) -> Result<(), String> {
         "https://api.telegram.org/bot{TELEGRAM_BOT_TOKEN}/sendMessage",
         &headers.to_string(),
         Some(&payload_bytes),
+        None,
     );
 
     match result {
@@ -914,15 +911,10 @@ fn handle_message(message: TelegramMessage) {
     let is_private = message.chat.chat_type == "private";
 
     // Owner validation: when owner_id is set, only that user can message
-    let owner_configured = channel_host::workspace_read(OWNER_ID_PATH)
-        .map(|s| !s.is_empty())
-        .unwrap_or(false);
+    let owner_id_str = channel_host::workspace_read(OWNER_ID_PATH).filter(|s| !s.is_empty());
 
-    if owner_configured {
-        if let Ok(owner_id) = channel_host::workspace_read(OWNER_ID_PATH)
-            .unwrap()
-            .parse::<i64>()
-        {
+    if let Some(ref id_str) = owner_id_str {
+        if let Ok(owner_id) = id_str.parse::<i64>() {
             if from.id != owner_id {
                 channel_host::log(
                     channel_host::LogLevel::Debug,
@@ -936,8 +928,8 @@ fn handle_message(message: TelegramMessage) {
         }
     } else if is_private {
         // No owner_id: apply dm_policy for private chats
-        let dm_policy = channel_host::workspace_read(DM_POLICY_PATH)
-            .unwrap_or_else(|| "pairing".to_string());
+        let dm_policy =
+            channel_host::workspace_read(DM_POLICY_PATH).unwrap_or_else(|| "pairing".to_string());
 
         if dm_policy != "open" {
             // Build effective allow list: config allow_from + pairing store
@@ -1000,8 +992,7 @@ fn handle_message(message: TelegramMessage) {
 
         if !respond_to_all {
             let has_command = content.starts_with('/');
-            let bot_username = channel_host::workspace_read(BOT_USERNAME_PATH)
-                .unwrap_or_default();
+            let bot_username = channel_host::workspace_read(BOT_USERNAME_PATH).unwrap_or_default();
             let has_bot_mention = if bot_username.is_empty() {
                 content.contains('@')
             } else {

channels-src/whatsapp/src/lib.rs

@@ -254,10 +254,19 @@ struct WhatsAppChannel;
 
 impl Guest for WhatsAppChannel {
     fn on_start(config_json: String) -> Result<ChannelConfig, String> {
-        let config: WhatsAppConfig = serde_json::from_str(&config_json).unwrap_or(WhatsAppConfig {
-            api_version: default_api_version(),
-            reply_to_message: default_reply_to_message(),
-        });
+        let config: WhatsAppConfig = match serde_json::from_str(&config_json) {
+            Ok(c) => c,
+            Err(e) => {
+                channel_host::log(
+                    channel_host::LogLevel::Warn,
+                    &format!("Failed to parse WhatsApp config, using defaults: {}", e),
+                );
+                WhatsAppConfig {
+                    api_version: default_api_version(),
+                    reply_to_message: default_reply_to_message(),
+                }
+            }
+        };
 
         channel_host::log(
             channel_host::LogLevel::Info,
@@ -267,6 +276,9 @@ impl Guest for WhatsAppChannel {
             ),
         );
 
+        // Persist api_version in workspace so on_respond() can read it
+        let _ = channel_host::workspace_write("channels/whatsapp/api_version", &config.api_version);
+
         // WhatsApp Cloud API is webhook-only, no polling available
         Ok(ChannelConfig {
             display_name: "WhatsApp".to_string(),
@@ -327,11 +339,16 @@ impl Guest for WhatsAppChannel {
         let metadata: WhatsAppMessageMetadata = serde_json::from_str(&response.metadata_json)
             .map_err(|e| format!("Failed to parse metadata: {}", e))?;
 
+        // Read api_version from workspace (set during on_start), fallback to default
+        let api_version = channel_host::workspace_read("channels/whatsapp/api_version")
+            .filter(|s| !s.is_empty())
+            .unwrap_or_else(|| "v18.0".to_string());
+
         // Build WhatsApp API URL with token placeholder
         // Host will replace {WHATSAPP_ACCESS_TOKEN} with actual token in Authorization header
         let api_url = format!(
-            "https://graph.facebook.com/v18.0/{}/messages",
-            metadata.phone_number_id
+            "https://graph.facebook.com/{}/{}/messages",
+            api_version, metadata.phone_number_id
         );
 
         // Build sendMessage payload

deploy/cloud-sql-proxy.service

@@ -0,0 +1,13 @@
+[Unit]
+Description=Cloud SQL Auth Proxy
+After=network.target
+
+[Service]
+Type=simple
+DynamicUser=yes
+ExecStart=/usr/local/bin/cloud-sql-proxy ironclaw-prod:us-central1:ironclaw-db --port=5432
+Restart=always
+RestartSec=5
+
+[Install]
+WantedBy=multi-user.target

deploy/env.example

@@ -0,0 +1,27 @@
+# WARNING: Replace all CHANGE_ME values before deploying.
+# Do not use placeholder passwords in production.
+DATABASE_URL=postgres://ironclaw:CHANGE_ME@localhost:5432/ironclaw
+
+# NEAR AI
+NEARAI_SESSION_TOKEN=CHANGE_ME
+NEARAI_MODEL=claude-3-5-sonnet-20241022
+NEARAI_BASE_URL=https://cloud-api.near.ai
+NEARAI_AUTH_URL=https://private.near.ai
+NEARAI_API_MODE=chat_completions
+
+# Agent
+AGENT_NAME=ironclaw
+CLI_ENABLED=false
+
+# Web Gateway
+GATEWAY_ENABLED=true
+# 0.0.0.0 binds to all interfaces (required for Docker --network=host).
+# Use 127.0.0.1 if running outside Docker or for local-only access.
+GATEWAY_HOST=0.0.0.0
+GATEWAY_PORT=3000
+GATEWAY_AUTH_TOKEN=CHANGE_ME
+
+# Disabled for initial deploy
+SANDBOX_ENABLED=false
+HEARTBEAT_ENABLED=false
+EMBEDDING_ENABLED=false

deploy/ironclaw.service

@@ -0,0 +1,20 @@
+[Unit]
+Description=IronClaw AI Assistant
+After=cloud-sql-proxy.service docker.service
+Requires=cloud-sql-proxy.service
+
+[Service]
+Type=simple
+ExecStartPre=/usr/bin/docker pull us-central1-docker.pkg.dev/ironclaw-prod/ironclaw/agent:latest
+ExecStart=/usr/bin/docker run --rm \
+  --name ironclaw \
+  --env-file /opt/ironclaw/.env \
+  --network=host \
+  us-central1-docker.pkg.dev/ironclaw-prod/ironclaw/agent:latest \
+  --no-onboard
+ExecStop=/usr/bin/docker stop ironclaw
+Restart=always
+RestartSec=10
+
+[Install]
+WantedBy=multi-user.target

deploy/setup.sh

@@ -0,0 +1,68 @@
+#!/usr/bin/env bash
+# VM bootstrap script for IronClaw on GCP Compute Engine.
+#
+# Run on a fresh Debian 12 VM after SSH:
+#   sudo bash setup.sh
+#
+# Prerequisites:
+#   - VM has the ironclaw-vm service account attached
+#   - Cloud SQL Auth Proxy accessible via IAM
+#   - Artifact Registry image pushed
+
+set -euo pipefail
+
+# Must run as root
+if [ "$(id -u)" -ne 0 ]; then
+  echo "ERROR: This script must be run as root (sudo bash setup.sh)"
+  exit 1
+fi
+
+echo "==> Installing Docker"
+apt-get update
+apt-get install -y docker.io
+systemctl enable docker
+systemctl start docker
+
+echo "==> Installing Cloud SQL Auth Proxy"
+curl -fsSL -o /usr/local/bin/cloud-sql-proxy \
+  https://storage.googleapis.com/cloud-sql-connectors/cloud-sql-proxy/v2.14.3/cloud-sql-proxy.linux.amd64
+chmod +x /usr/local/bin/cloud-sql-proxy
+
+echo "==> Installing systemd services"
+cp /tmp/deploy/cloud-sql-proxy.service /etc/systemd/system/
+cp /tmp/deploy/ironclaw.service /etc/systemd/system/
+systemctl daemon-reload
+
+echo "==> Starting Cloud SQL Auth Proxy"
+systemctl enable cloud-sql-proxy
+systemctl start cloud-sql-proxy
+
+echo "==> Configuring Docker registry auth"
+# The VM service account provides Artifact Registry access
+gcloud auth configure-docker us-central1-docker.pkg.dev --quiet
+
+echo "==> Creating config directory"
+# Owned by root, readable only by root. Docker reads --env-file as root
+# before dropping to uid 1000 (ironclaw) inside the container.
+mkdir -p /opt/ironclaw
+chmod 700 /opt/ironclaw
+
+if [ ! -f /opt/ironclaw/.env ]; then
+  echo "WARNING: /opt/ironclaw/.env does not exist."
+  echo "Create it with your configuration before starting IronClaw."
+  echo "See deploy/env.example for the required variables."
+  echo ""
+  echo "Then run: systemctl enable ironclaw && systemctl start ironclaw"
+else
+  chmod 600 /opt/ironclaw/.env
+  echo "==> Starting IronClaw"
+  systemctl enable ironclaw
+  systemctl start ironclaw
+fi
+
+echo "==> Setup complete"
+echo ""
+echo "Verify with:"
+echo "  systemctl status cloud-sql-proxy"
+echo "  systemctl status ironclaw"
+echo "  docker logs ironclaw"

docker-compose.yml

@@ -0,0 +1,20 @@
+# Local development only — do NOT use these credentials in production.
+services:
+  postgres:
+    image: pgvector/pgvector:pg16
+    ports:
+      - "5432:5432"
+    environment:
+      POSTGRES_DB: ironclaw
+      POSTGRES_USER: ironclaw
+      POSTGRES_PASSWORD: ironclaw  # dev-only, change for any non-local deployment
+    volumes:
+      - pgdata:/var/lib/postgresql/data
+    healthcheck:
+      test: ["CMD-SHELL", "pg_isready -U ironclaw"]
+      interval: 5s
+      timeout: 3s
+      retries: 5
+
+volumes:
+  pgdata:

examples/test_heartbeat.rs

@@ -81,7 +81,6 @@ async fn main() -> anyhow::Result<()> {
     let session = create_session_manager(SessionConfig {
         auth_base_url: config.llm.nearai.auth_base_url.clone(),
         session_path: config.llm.nearai.session_path.clone(),
-        ..Default::default()
     })
     .await;
     let llm = create_llm_provider(&config.llm, session)?;

src/agent/agent_loop.rs

@@ -19,16 +19,17 @@ use crate::channels::{ChannelManager, IncomingMessage, OutgoingResponse, StatusU
 use crate::config::{AgentConfig, HeartbeatConfig, RoutineConfig};
 use crate::context::ContextManager;
 use crate::context::JobContext;
+use crate::db::Database;
 use crate::error::Error;
 use crate::extensions::ExtensionManager;
-use crate::history::Store;
+use crate::hooks::HookRegistry;
 use crate::llm::{ChatMessage, LlmProvider, Reasoning, ReasoningContext, RespondResult};
 use crate::safety::SafetyLayer;
 use crate::tools::ToolRegistry;
 use crate::workspace::Workspace;
 
 /// Collapse a tool output string into a single-line preview for display.
-fn truncate_for_preview(output: &str, max_chars: usize) -> String {
+pub(crate) fn truncate_for_preview(output: &str, max_chars: usize) -> String {
     let collapsed: String = output
         .chars()
         .take(max_chars + 50)
@@ -37,8 +38,14 @@ fn truncate_for_preview(output: &str, max_chars: usize) -> String {
         .split_whitespace()
         .collect::<Vec<_>>()
         .join(" ");
-    if collapsed.len() > max_chars {
-        format!("{}...", &collapsed[..max_chars])
+    // char_indices gives us byte offsets at char boundaries, so the slice is always valid UTF-8.
+    if collapsed.chars().count() > max_chars {
+        let byte_offset = collapsed
+            .char_indices()
+            .nth(max_chars)
+            .map(|(i, _)| i)
+            .unwrap_or(collapsed.len());
+        format!("{}...", &collapsed[..byte_offset])
     } else {
         collapsed
     }
@@ -59,12 +66,16 @@ enum AgenticLoopResult {
 ///
 /// Bundles the shared components to reduce argument count.
 pub struct AgentDeps {
-    pub store: Option<Arc<Store>>,
+    pub store: Option<Arc<dyn Database>>,
     pub llm: Arc<dyn LlmProvider>,
+    /// Cheap/fast LLM for lightweight tasks (heartbeat, routing, evaluation).
+    /// Falls back to the main `llm` if None.
+    pub cheap_llm: Option<Arc<dyn LlmProvider>>,
     pub safety: Arc<SafetyLayer>,
     pub tools: Arc<ToolRegistry>,
     pub workspace: Option<Arc<Workspace>>,
     pub extension_manager: Option<Arc<ExtensionManager>>,
+    pub hooks: Arc<HookRegistry>,
 }
 
 /// The main agent that coordinates all components.
@@ -107,6 +118,7 @@ impl Agent {
             deps.safety.clone(),
             deps.tools.clone(),
             deps.store.clone(),
+            deps.hooks.clone(),
         ));
 
         Self {
@@ -124,7 +136,7 @@ impl Agent {
     }
 
     // Convenience accessors
-    fn store(&self) -> Option<&Arc<Store>> {
+    fn store(&self) -> Option<&Arc<dyn Database>> {
         self.deps.store.as_ref()
     }
 
@@ -132,6 +144,11 @@ impl Agent {
         &self.deps.llm
     }
 
+    /// Get the cheap/fast LLM provider, falling back to the main one.
+    fn cheap_llm(&self) -> &Arc<dyn LlmProvider> {
+        self.deps.cheap_llm.as_ref().unwrap_or(&self.deps.llm)
+    }
+
     fn safety(&self) -> &Arc<SafetyLayer> {
         &self.deps.safety
     }
@@ -144,6 +161,10 @@ impl Agent {
         self.deps.workspace.as_ref()
     }
 
+    fn hooks(&self) -> &Arc<HookRegistry> {
+        &self.deps.hooks
+    }
+
     /// Run the agent main loop.
     pub async fn run(self) -> Result<(), Error> {
         // Start channels
@@ -295,7 +316,7 @@ impl Agent {
                     Some(spawn_heartbeat(
                         config,
                         workspace.clone(),
-                        self.llm().clone(),
+                        self.cheap_llm().clone(),
                         Some(notify_tx),
                     ))
                 } else {
@@ -411,10 +432,32 @@ impl Agent {
 
             match self.handle_message(&message).await {
                 Ok(Some(response)) if !response.is_empty() => {
-                    let _ = self
-                        .channels
-                        .respond(&message, OutgoingResponse::text(response))
-                        .await;
+                    // Hook: BeforeOutbound — allow hooks to modify or suppress outbound
+                    let event = crate::hooks::HookEvent::Outbound {
+                        user_id: message.user_id.clone(),
+                        channel: message.channel.clone(),
+                        content: response.clone(),
+                        thread_id: message.thread_id.clone(),
+                    };
+                    match self.hooks().run(&event).await {
+                        Err(err) => {
+                            tracing::warn!("BeforeOutbound hook blocked response: {}", err);
+                        }
+                        Ok(crate::hooks::HookOutcome::Continue {
+                            modified: Some(new_content),
+                        }) => {
+                            let _ = self
+                                .channels
+                                .respond(&message, OutgoingResponse::text(new_content))
+                                .await;
+                        }
+                        _ => {
+                            let _ = self
+                                .channels
+                                .respond(&message, OutgoingResponse::text(response))
+                                .await;
+                        }
+                    }
                 }
                 Ok(Some(_)) => {
                     // Empty response, nothing to send (e.g. approval handled via send_status)
@@ -460,7 +503,33 @@ impl Agent {
 
     async fn handle_message(&self, message: &IncomingMessage) -> Result<Option<String>, Error> {
         // Parse submission type first
-        let submission = SubmissionParser::parse(&message.content);
+        let mut submission = SubmissionParser::parse(&message.content);
+
+        // Hook: BeforeInbound — allow hooks to modify or reject user input
+        if let Submission::UserInput { ref content } = submission {
+            let event = crate::hooks::HookEvent::Inbound {
+                user_id: message.user_id.clone(),
+                channel: message.channel.clone(),
+                content: content.clone(),
+                thread_id: message.thread_id.clone(),
+            };
+            match self.hooks().run(&event).await {
+                Err(crate::hooks::HookError::Rejected { reason }) => {
+                    return Ok(Some(format!("[Message rejected: {}]", reason)));
+                }
+                Err(err) => {
+                    return Ok(Some(format!("[Message blocked by hook policy: {}]", err)));
+                }
+                Ok(crate::hooks::HookOutcome::Continue {
+                    modified: Some(new_content),
+                }) => {
+                    submission = Submission::UserInput {
+                        content: new_content,
+                    };
+                }
+                _ => {} // Continue, fail-open errors already logged in registry
+            }
+        }
 
         // Hydrate thread from DB if it's a historical thread not in memory
         if let Some(ref external_thread_id) = message.thread_id {
@@ -654,19 +723,17 @@ impl Agent {
         }
 
         // Restore response chain from conversation metadata
-        if let Some(store) = self.store() {
-            if let Ok(Some(metadata)) = store.get_conversation_metadata(thread_uuid).await {
-                if let Some(rid) = metadata
-                    .get("last_response_id")
-                    .and_then(|v| v.as_str())
-                    .map(String::from)
-                {
-                    thread.last_response_id = Some(rid.clone());
-                    self.llm()
-                        .seed_response_chain(&thread_uuid.to_string(), rid);
-                    tracing::debug!("Restored response chain for thread {}", thread_uuid);
-                }
-            }
+        if let Some(store) = self.store()
+            && let Ok(Some(metadata)) = store.get_conversation_metadata(thread_uuid).await
+            && let Some(rid) = metadata
+                .get("last_response_id")
+                .and_then(|v| v.as_str())
+                .map(String::from)
+        {
+            thread.last_response_id = Some(rid.clone());
+            self.llm()
+                .seed_response_chain(&thread_uuid.to_string(), rid);
+            tracing::debug!("Restored response chain for thread {}", thread_uuid);
         }
 
         // Insert into session and register with session manager
@@ -871,6 +938,27 @@ impl Agent {
         // Complete, fail, or request approval
         match result {
             Ok(AgenticLoopResult::Response(response)) => {
+                // Hook: TransformResponse — allow hooks to modify or reject the final response
+                let response = {
+                    let event = crate::hooks::HookEvent::ResponseTransform {
+                        user_id: message.user_id.clone(),
+                        thread_id: thread_id.to_string(),
+                        response: response.clone(),
+                    };
+                    match self.hooks().run(&event).await {
+                        Err(crate::hooks::HookError::Rejected { reason }) => {
+                            format!("[Response filtered: {}]", reason)
+                        }
+                        Err(err) => {
+                            format!("[Response blocked by hook policy: {}]", err)
+                        }
+                        Ok(crate::hooks::HookOutcome::Continue {
+                            modified: Some(new_response),
+                        }) => new_response,
+                        _ => response, // fail-open: use original
+                    }
+                };
+
                 thread.complete_turn(&response);
                 self.persist_response_chain(thread);
                 let _ = self
@@ -954,13 +1042,12 @@ impl Agent {
                 return;
             }
 
-            if let Some(ref resp) = response {
-                if let Err(e) = store
+            if let Some(ref resp) = response
+                && let Err(e) = store
                     .add_conversation_message(thread_id, "assistant", resp)
                     .await
-                {
-                    tracing::warn!("Failed to persist assistant message: {}", e);
-                }
+            {
+                tracing::warn!("Failed to persist assistant message: {}", e);
             }
         });
     }
@@ -1058,14 +1145,14 @@ impl Agent {
             // Check if interrupted
             {
                 let sess = session.lock().await;
-                if let Some(thread) = sess.threads.get(&thread_id) {
-                    if thread.state == ThreadState::Interrupted {
-                        return Err(crate::error::JobError::ContextError {
-                            id: thread_id,
-                            reason: "Interrupted".to_string(),
-                        }
-                        .into());
+                if let Some(thread) = sess.threads.get(&thread_id)
+                    && thread.state == ThreadState::Interrupted
+                {
+                    return Err(crate::error::JobError::ContextError {
+                        id: thread_id,
+                        reason: "Interrupted".to_string(),
                     }
+                    .into());
                 }
             }
 
@@ -1082,9 +1169,16 @@ impl Agent {
                     m
                 });
 
-            let result = reasoning.respond_with_tools(&context).await?;
+            let output = reasoning.respond_with_tools(&context).await?;
 
-            match result {
+            // Track token usage for budget enforcement
+            tracing::debug!(
+                "LLM call used {} input + {} output tokens",
+                output.usage.input_tokens,
+                output.usage.output_tokens
+            );
+
+            match output.result {
                 RespondResult::Text(text) => {
                     // If no tools have been executed yet, prompt the LLM to use tools
                     // This handles the case where the model explains what it will do
@@ -1133,39 +1227,90 @@ impl Agent {
                     // Record tool calls in the thread
                     {
                         let mut sess = session.lock().await;
-                        if let Some(thread) = sess.threads.get_mut(&thread_id) {
-                            if let Some(turn) = thread.last_turn_mut() {
-                                for tc in &tool_calls {
-                                    turn.record_tool_call(&tc.name, tc.arguments.clone());
-                                }
+                        if let Some(thread) = sess.threads.get_mut(&thread_id)
+                            && let Some(turn) = thread.last_turn_mut()
+                        {
+                            for tc in &tool_calls {
+                                turn.record_tool_call(&tc.name, tc.arguments.clone());
                             }
                         }
                     }
 
-                    // Execute each tool (with approval checking)
-                    for tc in tool_calls {
+                    // Execute each tool (with approval checking and hook interception)
+                    for mut tc in tool_calls {
                         // Check if tool requires approval
-                        if let Some(tool) = self.tools().get(&tc.name).await {
-                            if tool.requires_approval() {
-                                // Check if auto-approved for this session
-                                let is_auto_approved = {
-                                    let sess = session.lock().await;
-                                    sess.is_tool_auto_approved(&tc.name)
+                        if let Some(tool) = self.tools().get(&tc.name).await
+                            && tool.requires_approval()
+                        {
+                            // Check if auto-approved for this session
+                            let mut is_auto_approved = {
+                                let sess = session.lock().await;
+                                sess.is_tool_auto_approved(&tc.name)
+                            };
+
+                            // Let the tool inspect the specific parameters and
+                            // override auto-approval (e.g. destructive shell commands).
+                            if is_auto_approved && tool.requires_approval_for(&tc.arguments) {
+                                tracing::info!(
+                                    tool = %tc.name,
+                                    "Tool requires explicit approval for these parameters despite auto-approve"
+                                );
+                                is_auto_approved = false;
+                            }
+
+                            if !is_auto_approved {
+                                // Need approval - store pending request and return
+                                let pending = PendingApproval {
+                                    request_id: Uuid::new_v4(),
+                                    tool_name: tc.name.clone(),
+                                    parameters: tc.arguments.clone(),
+                                    description: tool.description().to_string(),
+                                    tool_call_id: tc.id.clone(),
+                                    context_messages: context_messages.clone(),
                                 };
 
-                                if !is_auto_approved {
-                                    // Need approval - store pending request and return
-                                    let pending = PendingApproval {
-                                        request_id: Uuid::new_v4(),
-                                        tool_name: tc.name.clone(),
-                                        parameters: tc.arguments.clone(),
-                                        description: tool.description().to_string(),
-                                        tool_call_id: tc.id.clone(),
-                                        context_messages: context_messages.clone(),
-                                    };
+                                return Ok(AgenticLoopResult::NeedApproval { pending });
+                            }
+                        }
 
-                                    return Ok(AgenticLoopResult::NeedApproval { pending });
+                        // Hook: BeforeToolCall — allow hooks to modify or reject tool calls
+                        {
+                            let event = crate::hooks::HookEvent::ToolCall {
+                                tool_name: tc.name.clone(),
+                                parameters: tc.arguments.clone(),
+                                user_id: message.user_id.clone(),
+                                context: "chat".to_string(),
+                            };
+                            match self.hooks().run(&event).await {
+                                Err(crate::hooks::HookError::Rejected { reason }) => {
+                                    context_messages.push(ChatMessage::tool_result(
+                                        &tc.id,
+                                        &tc.name,
+                                        format!("Tool call rejected by hook: {}", reason),
+                                    ));
+                                    continue;
                                 }
+                                Err(err) => {
+                                    context_messages.push(ChatMessage::tool_result(
+                                        &tc.id,
+                                        &tc.name,
+                                        format!("Tool call blocked by hook policy: {}", err),
+                                    ));
+                                    continue;
+                                }
+                                Ok(crate::hooks::HookOutcome::Continue {
+                                    modified: Some(new_params),
+                                }) => match serde_json::from_str(&new_params) {
+                                    Ok(parsed) => tc.arguments = parsed,
+                                    Err(e) => {
+                                        tracing::warn!(
+                                            tool = %tc.name,
+                                            "Hook returned non-JSON modification for ToolCall, ignoring: {}",
+                                            e
+                                        );
+                                    }
+                                },
+                                _ => {} // Continue, fail-open errors already logged
                             }
                         }
 
@@ -1196,34 +1341,34 @@ impl Agent {
                             )
                             .await;
 
-                        if let Ok(ref output) = tool_result {
-                            if !output.is_empty() {
-                                let _ = self
-                                    .channels
-                                    .send_status(
-                                        &message.channel,
-                                        StatusUpdate::ToolResult {
-                                            name: tc.name.clone(),
-                                            preview: truncate_for_preview(output, 200),
-                                        },
-                                        &message.metadata,
-                                    )
-                                    .await;
-                            }
+                        if let Ok(ref output) = tool_result
+                            && !output.is_empty()
+                        {
+                            let _ = self
+                                .channels
+                                .send_status(
+                                    &message.channel,
+                                    StatusUpdate::ToolResult {
+                                        name: tc.name.clone(),
+                                        preview: output.clone(),
+                                    },
+                                    &message.metadata,
+                                )
+                                .await;
                         }
 
                         // Record result in thread
                         {
                             let mut sess = session.lock().await;
-                            if let Some(thread) = sess.threads.get_mut(&thread_id) {
-                                if let Some(turn) = thread.last_turn_mut() {
-                                    match &tool_result {
-                                        Ok(output) => {
-                                            turn.record_tool_result(serde_json::json!(output));
-                                        }
-                                        Err(e) => {
-                                            turn.record_tool_error(e.to_string());
-                                        }
+                            if let Some(thread) = sess.threads.get_mut(&thread_id)
+                                && let Some(turn) = thread.last_turn_mut()
+                            {
+                                match &tool_result {
+                                    Ok(output) => {
+                                        turn.record_tool_result(serde_json::json!(output));
+                                    }
+                                    Err(e) => {
+                                        turn.record_tool_error(e.to_string());
                                     }
                                 }
                             }
@@ -1434,6 +1579,9 @@ impl Agent {
         session: Arc<Mutex<Session>>,
         thread_id: Uuid,
     ) -> Result<SubmissionResult, Error> {
+        // Lock session first, then undo manager -- consistent with process_user_input
+        // to avoid potential deadlocks.
+        let mut sess = session.lock().await;
         let undo_mgr = self.session_manager.get_undo_manager(thread_id).await;
         let mut mgr = undo_mgr.lock().await;
 
@@ -1441,7 +1589,6 @@ impl Agent {
             return Ok(SubmissionResult::ok_with_message("Nothing to undo."));
         }
 
-        let mut sess = session.lock().await;
         let thread = sess
             .threads
             .get_mut(&thread_id)
@@ -1452,12 +1599,10 @@ impl Agent {
         let current_turn = thread.turn_number();
 
         if let Some(checkpoint) = mgr.undo(current_turn, current_messages) {
-            // Extract values before consuming the reference
             let turn_number = checkpoint.turn_number;
-            let messages = checkpoint.messages.clone();
             let undo_count = mgr.undo_count();
             // Restore thread from checkpoint
-            thread.restore_from_messages(messages);
+            thread.restore_from_messages(checkpoint.messages);
             Ok(SubmissionResult::ok_with_message(format!(
                 "Undone to turn {}. {} undo(s) remaining.",
                 turn_number, undo_count
@@ -1472,6 +1617,9 @@ impl Agent {
         session: Arc<Mutex<Session>>,
         thread_id: Uuid,
     ) -> Result<SubmissionResult, Error> {
+        // Lock session first, then undo manager -- consistent with process_user_input
+        // to avoid potential deadlocks.
+        let mut sess = session.lock().await;
         let undo_mgr = self.session_manager.get_undo_manager(thread_id).await;
         let mut mgr = undo_mgr.lock().await;
 
@@ -1479,12 +1627,15 @@ impl Agent {
             return Ok(SubmissionResult::ok_with_message("Nothing to redo."));
         }
 
-        if let Some(checkpoint) = mgr.redo() {
-            let mut sess = session.lock().await;
-            let thread = sess
-                .threads
-                .get_mut(&thread_id)
-                .ok_or_else(|| Error::from(crate::error::JobError::NotFound { id: thread_id }))?;
+        let thread = sess
+            .threads
+            .get_mut(&thread_id)
+            .ok_or_else(|| Error::from(crate::error::JobError::NotFound { id: thread_id }))?;
+
+        let current_messages = thread.messages();
+        let current_turn = thread.turn_number();
+
+        if let Some(checkpoint) = mgr.redo(current_turn, current_messages) {
             thread.restore_from_messages(checkpoint.messages);
             Ok(SubmissionResult::ok_with_message(format!(
                 "Redone to turn {}.",
@@ -1606,17 +1757,17 @@ impl Agent {
         };
 
         // Verify request ID if provided
-        if let Some(req_id) = request_id {
-            if req_id != pending.request_id {
-                // Put it back and return error
-                let mut sess = session.lock().await;
-                if let Some(thread) = sess.threads.get_mut(&thread_id) {
-                    thread.await_approval(pending);
-                }
-                return Ok(SubmissionResult::error(
-                    "Request ID mismatch. Use the correct request ID.",
-                ));
+        if let Some(req_id) = request_id
+            && req_id != pending.request_id
+        {
+            // Put it back and return error
+            let mut sess = session.lock().await;
+            if let Some(thread) = sess.threads.get_mut(&thread_id) {
+                thread.await_approval(pending);
             }
+            return Ok(SubmissionResult::error(
+                "Request ID mismatch. Use the correct request ID.",
+            ));
         }
 
         if approved {
@@ -1670,20 +1821,20 @@ impl Agent {
                 )
                 .await;
 
-            if let Ok(ref output) = tool_result {
-                if !output.is_empty() {
-                    let _ = self
-                        .channels
-                        .send_status(
-                            &message.channel,
-                            StatusUpdate::ToolResult {
-                                name: pending.tool_name.clone(),
-                                preview: truncate_for_preview(output, 200),
-                            },
-                            &message.metadata,
-                        )
-                        .await;
-                }
+            if let Ok(ref output) = tool_result
+                && !output.is_empty()
+            {
+                let _ = self
+                    .channels
+                    .send_status(
+                        &message.channel,
+                        StatusUpdate::ToolResult {
+                            name: pending.tool_name.clone(),
+                            preview: output.clone(),
+                        },
+                        &message.metadata,
+                    )
+                    .await;
             }
 
             // Build context including the tool result
@@ -1692,15 +1843,15 @@ impl Agent {
             // Record result in thread
             {
                 let mut sess = session.lock().await;
-                if let Some(thread) = sess.threads.get_mut(&thread_id) {
-                    if let Some(turn) = thread.last_turn_mut() {
-                        match &tool_result {
-                            Ok(output) => {
-                                turn.record_tool_result(serde_json::json!(output));
-                            }
-                            Err(e) => {
-                                turn.record_tool_error(e.to_string());
-                            }
+                if let Some(thread) = sess.threads.get_mut(&thread_id)
+                    && let Some(turn) = thread.last_turn_mut()
+                {
+                    match &tool_result {
+                        Ok(output) => {
+                            turn.record_tool_result(serde_json::json!(output));
+                        }
+                        Err(e) => {
+                            turn.record_tool_error(e.to_string());
                         }
                     }
                 }
@@ -2060,15 +2211,15 @@ impl Agent {
         }
 
         // Persist new job to database (fire-and-forget)
-        if let Some(store) = self.store() {
-            if let Ok(ctx) = self.context_manager.get_context(job_id).await {
-                let store = store.clone();
-                tokio::spawn(async move {
-                    if let Err(e) = store.save_job(&ctx).await {
-                        tracing::warn!("Failed to persist new job {}: {}", job_id, e);
-                    }
-                });
-            }
+        if let Some(store) = self.store()
+            && let Ok(ctx) = self.context_manager.get_context(job_id).await
+        {
+            let store = store.clone();
+            tokio::spawn(async move {
+                if let Err(e) = store.save_job(&ctx).await {
+                    tracing::warn!("Failed to persist new job {}: {}", job_id, e);
+                }
+            });
         }
 
         // Schedule for execution
@@ -2148,10 +2299,10 @@ impl Agent {
 
         let mut output = String::from("Jobs:\n");
         for job_id in jobs {
-            if let Ok(ctx) = self.context_manager.get_context(job_id).await {
-                if ctx.user_id == user_id {
-                    output.push_str(&format!("  {} - {} ({:?})\n", job_id, ctx.title, ctx.state));
-                }
+            if let Ok(ctx) = self.context_manager.get_context(job_id).await
+                && ctx.user_id == user_id
+            {
+                output.push_str(&format!("  {} - {} ({:?})\n", job_id, ctx.title, ctx.state));
             }
         }
 
@@ -2602,4 +2753,70 @@ mod tests {
 
         assert!(detect_auth_awaiting("tool_activate", &result).is_none());
     }
+
+    // --- truncate_for_preview tests ---
+
+    use super::truncate_for_preview;
+
+    #[test]
+    fn test_truncate_short_input() {
+        assert_eq!(truncate_for_preview("hello", 10), "hello");
+    }
+
+    #[test]
+    fn test_truncate_empty_input() {
+        assert_eq!(truncate_for_preview("", 10), "");
+    }
+
+    #[test]
+    fn test_truncate_exact_length() {
+        assert_eq!(truncate_for_preview("hello", 5), "hello");
+    }
+
+    #[test]
+    fn test_truncate_over_limit() {
+        let result = truncate_for_preview("hello world, this is long", 10);
+        assert!(result.ends_with("..."));
+        // "hello worl" = 10 chars + "..."
+        assert_eq!(result, "hello worl...");
+    }
+
+    #[test]
+    fn test_truncate_collapses_newlines() {
+        let result = truncate_for_preview("line1\nline2\nline3", 100);
+        assert!(!result.contains('\n'));
+        assert_eq!(result, "line1 line2 line3");
+    }
+
+    #[test]
+    fn test_truncate_collapses_whitespace() {
+        let result = truncate_for_preview("hello   world", 100);
+        assert_eq!(result, "hello world");
+    }
+
+    #[test]
+    fn test_truncate_multibyte_utf8() {
+        // Each emoji is 4 bytes. Truncating at char boundary must not panic.
+        let input = "😀😁😂🤣😃😄😅😆😉😊";
+        let result = truncate_for_preview(input, 5);
+        assert!(result.ends_with("..."));
+        // First 5 chars = 5 emoji
+        assert_eq!(result, "😀😁😂🤣😃...");
+    }
+
+    #[test]
+    fn test_truncate_cjk_characters() {
+        // CJK chars are 3 bytes each in UTF-8.
+        let input = "你好世界测试数据很长的字符串";
+        let result = truncate_for_preview(input, 4);
+        assert_eq!(result, "你好世界...");
+    }
+
+    #[test]
+    fn test_truncate_mixed_multibyte_and_ascii() {
+        let input = "hello 世界 foo";
+        let result = truncate_for_preview(input, 8);
+        // 'h','e','l','l','o',' ','世','界' = 8 chars
+        assert_eq!(result, "hello 世界...");
+    }
 }

src/agent/mod.rs

@@ -26,6 +26,7 @@ pub mod task;
 pub mod undo;
 pub mod worker;
 
+pub(crate) use agent_loop::truncate_for_preview;
 pub use agent_loop::{Agent, AgentDeps};
 pub use compaction::{CompactionResult, ContextCompactor};
 pub use context_monitor::{CompactionStrategy, ContextBreakdown, ContextMonitor};

src/agent/routine_engine.rs

@@ -11,6 +11,7 @@
 //! Full-job routines are delegated to the existing `Scheduler`.
 
 use std::sync::Arc;
+use std::sync::atomic::{AtomicUsize, Ordering};
 use std::time::Duration;
 
 use chrono::Utc;
@@ -23,20 +24,20 @@ use crate::agent::routine::{
 };
 use crate::channels::{IncomingMessage, OutgoingResponse};
 use crate::config::RoutineConfig;
-use crate::history::Store;
+use crate::db::Database;
 use crate::llm::{ChatMessage, CompletionRequest, FinishReason, LlmProvider};
 use crate::workspace::Workspace;
 
 /// The routine execution engine.
 pub struct RoutineEngine {
     config: RoutineConfig,
-    store: Arc<Store>,
+    store: Arc<dyn Database>,
     llm: Arc<dyn LlmProvider>,
     workspace: Arc<Workspace>,
     /// Sender for notifications (routed to channel manager).
     notify_tx: mpsc::Sender<OutgoingResponse>,
     /// Currently running routine count (across all routines).
-    running_count: Arc<RwLock<usize>>,
+    running_count: Arc<AtomicUsize>,
     /// Compiled event regex cache: routine_id -> compiled regex.
     event_cache: Arc<RwLock<Vec<(Uuid, Routine, Regex)>>>,
 }
@@ -44,7 +45,7 @@ pub struct RoutineEngine {
 impl RoutineEngine {
     pub fn new(
         config: RoutineConfig,
-        store: Arc<Store>,
+        store: Arc<dyn Database>,
         llm: Arc<dyn LlmProvider>,
         workspace: Arc<Workspace>,
         notify_tx: mpsc::Sender<OutgoingResponse>,
@@ -55,7 +56,7 @@ impl RoutineEngine {
             llm,
             workspace,
             notify_tx,
-            running_count: Arc::new(RwLock::new(0)),
+            running_count: Arc::new(AtomicUsize::new(0)),
             event_cache: Arc::new(RwLock::new(Vec::new())),
         }
     }
@@ -102,10 +103,9 @@ impl RoutineEngine {
             if let Trigger::Event {
                 channel: Some(ch), ..
             } = &routine.trigger
+                && ch != &message.channel
             {
-                if ch != &message.channel {
-                    continue;
-                }
+                continue;
             }
 
             // Regex match
@@ -126,7 +126,7 @@ impl RoutineEngine {
             }
 
             // Global capacity check
-            if *self.running_count.read().await >= self.config.max_concurrent_routines {
+            if self.running_count.load(Ordering::Relaxed) >= self.config.max_concurrent_routines {
                 tracing::warn!(routine = %routine.name, "Skipped: global max concurrent reached");
                 continue;
             }
@@ -150,7 +150,7 @@ impl RoutineEngine {
         };
 
         for routine in routines {
-            if *self.running_count.read().await >= self.config.max_concurrent_routines {
+            if self.running_count.load(Ordering::Relaxed) >= self.config.max_concurrent_routines {
                 tracing::warn!("Global max concurrent routines reached, skipping remaining");
                 break;
             }
@@ -293,21 +293,18 @@ impl RoutineEngine {
 
 /// Shared context passed to the execution function.
 struct EngineContext {
-    store: Arc<Store>,
+    store: Arc<dyn Database>,
     llm: Arc<dyn LlmProvider>,
     workspace: Arc<Workspace>,
     notify_tx: mpsc::Sender<OutgoingResponse>,
-    running_count: Arc<RwLock<usize>>,
+    running_count: Arc<AtomicUsize>,
     max_lightweight_tokens: u32,
 }
 
 /// Execute a routine run. Handles both lightweight and full_job modes.
 async fn execute_routine(ctx: EngineContext, routine: Routine, run: RoutineRun) {
-    // Increment running count
-    {
-        let mut count = ctx.running_count.write().await;
-        *count += 1;
-    }
+    // Increment running count (atomic: survives panics in the execution below)
+    ctx.running_count.fetch_add(1, Ordering::Relaxed);
 
     let result = match &routine.action {
         RoutineAction::Lightweight {
@@ -327,10 +324,7 @@ async fn execute_routine(ctx: EngineContext, routine: Routine, run: RoutineRun)
     };
 
     // Decrement running count
-    {
-        let mut count = ctx.running_count.write().await;
-        *count = count.saturating_sub(1);
-    }
+    ctx.running_count.fetch_sub(1, Ordering::Relaxed);
 
     // Process result
     let (status, summary, tokens) = match result {
@@ -568,7 +562,8 @@ fn truncate(s: &str, max: usize) -> String {
     if s.len() <= max {
         s.to_string()
     } else {
-        format!("{}...", &s[..max])
+        let end = crate::util::floor_char_boundary(s, max);
+        format!("{}...", &s[..end])
     }
 }
 

src/agent/scheduler.rs

@@ -12,8 +12,9 @@ use crate::agent::task::{Task, TaskContext, TaskOutput};
 use crate::agent::worker::{Worker, WorkerDeps};
 use crate::config::AgentConfig;
 use crate::context::{ContextManager, JobContext, JobState};
+use crate::db::Database;
 use crate::error::{Error, JobError};
-use crate::history::Store;
+use crate::hooks::HookRegistry;
 use crate::llm::LlmProvider;
 use crate::safety::SafetyLayer;
 use crate::tools::ToolRegistry;
@@ -48,7 +49,8 @@ pub struct Scheduler {
     llm: Arc<dyn LlmProvider>,
     safety: Arc<SafetyLayer>,
     tools: Arc<ToolRegistry>,
-    store: Option<Arc<Store>>,
+    store: Option<Arc<dyn Database>>,
+    hooks: Arc<HookRegistry>,
     /// Running jobs (main LLM-driven jobs).
     jobs: Arc<RwLock<HashMap<Uuid, ScheduledJob>>>,
     /// Running sub-tasks (tool executions, background tasks).
@@ -63,7 +65,8 @@ impl Scheduler {
         llm: Arc<dyn LlmProvider>,
         safety: Arc<SafetyLayer>,
         tools: Arc<ToolRegistry>,
-        store: Option<Arc<Store>>,
+        store: Option<Arc<dyn Database>>,
+        hooks: Arc<HookRegistry>,
     ) -> Self {
         Self {
             config,
@@ -72,6 +75,7 @@ impl Scheduler {
             safety,
             tools,
             store,
+            hooks,
             jobs: Arc::new(RwLock::new(HashMap::new())),
             subtasks: Arc::new(RwLock::new(HashMap::new())),
         }
@@ -79,63 +83,64 @@ impl Scheduler {
 
     /// Schedule a job for execution.
     pub async fn schedule(&self, job_id: Uuid) -> Result<(), JobError> {
-        // Check if already scheduled
-        if self.jobs.read().await.contains_key(&job_id) {
-            return Ok(());
-        }
+        // Hold write lock for the entire check-insert sequence to prevent
+        // TOCTOU races where two concurrent calls both pass the checks.
+        {
+            let mut jobs = self.jobs.write().await;
 
-        // Check capacity
-        let current_count = self.jobs.read().await.len();
-        if current_count >= self.config.max_parallel_jobs {
-            return Err(JobError::MaxJobsExceeded {
-                max: self.config.max_parallel_jobs,
-            });
-        }
-
-        // Transition job to in_progress
-        self.context_manager
-            .update_context(job_id, |ctx| {
-                ctx.transition_to(
-                    JobState::InProgress,
-                    Some("Scheduled for execution".to_string()),
-                )
-            })
-            .await?
-            .map_err(|s| JobError::ContextError {
-                id: job_id,
-                reason: s,
-            })?;
-
-        // Create worker channel
-        let (tx, rx) = mpsc::channel(16);
-
-        // Create worker with shared dependencies
-        let deps = WorkerDeps {
-            context_manager: self.context_manager.clone(),
-            llm: self.llm.clone(),
-            safety: self.safety.clone(),
-            tools: self.tools.clone(),
-            store: self.store.clone(),
-            timeout: self.config.job_timeout,
-            use_planning: self.config.use_planning,
-        };
-        let worker = Worker::new(job_id, deps);
-
-        // Spawn worker task
-        let handle = tokio::spawn(async move {
-            if let Err(e) = worker.run(rx).await {
-                tracing::error!("Worker for job {} failed: {}", job_id, e);
+            if jobs.contains_key(&job_id) {
+                return Ok(());
             }
-        });
 
-        // Start the worker
-        let _ = tx.send(WorkerMessage::Start).await;
+            if jobs.len() >= self.config.max_parallel_jobs {
+                return Err(JobError::MaxJobsExceeded {
+                    max: self.config.max_parallel_jobs,
+                });
+            }
 
-        // Store the scheduled job
-        self.jobs
-            .write()
-            .await
-            .insert(job_id, ScheduledJob { handle, tx });
+            // Transition job to in_progress
+            self.context_manager
+                .update_context(job_id, |ctx| {
+                    ctx.transition_to(
+                        JobState::InProgress,
+                        Some("Scheduled for execution".to_string()),
+                    )
+                })
+                .await?
+                .map_err(|s| JobError::ContextError {
+                    id: job_id,
+                    reason: s,
+                })?;
+
+            // Create worker channel
+            let (tx, rx) = mpsc::channel(16);
+
+            // Create worker with shared dependencies
+            let deps = WorkerDeps {
+                context_manager: self.context_manager.clone(),
+                llm: self.llm.clone(),
+                safety: self.safety.clone(),
+                tools: self.tools.clone(),
+                store: self.store.clone(),
+                hooks: self.hooks.clone(),
+                timeout: self.config.job_timeout,
+                use_planning: self.config.use_planning,
+            };
+            let worker = Worker::new(job_id, deps);
+
+            // Spawn worker task
+            let handle = tokio::spawn(async move {
+                if let Err(e) = worker.run(rx).await {
+                    tracing::error!("Worker for job {} failed: {}", job_id, e);
+                }
+            });
+
+            // Start the worker
+            let _ = tx.send(WorkerMessage::Start).await;
+
+            // Insert while still holding the write lock
+            jobs.insert(job_id, ScheduledJob { handle, tx });
+        }
 
         // Cleanup task for this job to avoid capacity leaks
         let jobs = Arc::clone(&self.jobs);

src/agent/self_repair.rs

@@ -8,8 +8,8 @@ use chrono::{DateTime, Utc};
 use uuid::Uuid;
 
 use crate::context::{ContextManager, JobState};
+use crate::db::Database;
 use crate::error::RepairError;
-use crate::history::Store;
 use crate::tools::{BuildRequirement, Language, SoftwareBuilder, SoftwareType, ToolRegistry};
 
 /// A job that has been detected as stuck.
@@ -69,7 +69,7 @@ pub struct DefaultSelfRepair {
     #[allow(dead_code)] // Will be used for time-based stuck detection
     stuck_threshold: Duration,
     max_repair_attempts: u32,
-    store: Option<Arc<Store>>,
+    store: Option<Arc<dyn Database>>,
     builder: Option<Arc<dyn SoftwareBuilder>>,
     #[allow(dead_code)] // Will be used for tool hot-reload after repair
     tools: Option<Arc<ToolRegistry>>,
@@ -94,7 +94,7 @@ impl DefaultSelfRepair {
 
     /// Add a Store for tool failure tracking.
     #[allow(dead_code)] // Public API for configuring repair with persistence
-    pub fn with_store(mut self, store: Arc<Store>) -> Self {
+    pub fn with_store(mut self, store: Arc<dyn Database>) -> Self {
         self.store = Some(store);
         self
     }
@@ -119,25 +119,25 @@ impl SelfRepair for DefaultSelfRepair {
         let mut stuck_jobs = Vec::new();
 
         for job_id in stuck_ids {
-            if let Ok(ctx) = self.context_manager.get_context(job_id).await {
-                if ctx.state == JobState::Stuck {
-                    let stuck_duration = ctx
-                        .started_at
-                        .map(|start| {
-                            let now = Utc::now();
-                            let duration = now.signed_duration_since(start);
-                            Duration::from_secs(duration.num_seconds().max(0) as u64)
-                        })
-                        .unwrap_or_default();
+            if let Ok(ctx) = self.context_manager.get_context(job_id).await
+                && ctx.state == JobState::Stuck
+            {
+                let stuck_duration = ctx
+                    .started_at
+                    .map(|start| {
+                        let now = Utc::now();
+                        let duration = now.signed_duration_since(start);
+                        Duration::from_secs(duration.num_seconds().max(0) as u64)
+                    })
+                    .unwrap_or_default();
 
-                    stuck_jobs.push(StuckJob {
-                        job_id,
-                        last_activity: ctx.started_at.unwrap_or(ctx.created_at),
-                        stuck_duration,
-                        last_error: None,
-                        repair_attempts: ctx.repair_attempts,
-                    });
-                }
+                stuck_jobs.push(StuckJob {
+                    job_id,
+                    last_activity: ctx.started_at.unwrap_or(ctx.created_at),
+                    stuck_duration,
+                    last_error: None,
+                    repair_attempts: ctx.repair_attempts,
+                });
             }
         }
 

src/agent/session.rs

@@ -346,11 +346,11 @@ impl Thread {
                 let mut turn = Turn::new(turn_number, &msg.content);
 
                 // Check if next is assistant response
-                if let Some(next) = iter.peek() {
-                    if next.role == crate::llm::Role::Assistant {
-                        let response = iter.next().expect("peeked");
-                        turn.complete(&response.content);
-                    }
+                if let Some(next) = iter.peek()
+                    && next.role == crate::llm::Role::Assistant
+                {
+                    let response = iter.next().expect("peeked");
+                    turn.complete(&response.content);
                 }
 
                 self.turns.push(turn);

src/agent/session_manager.rs

@@ -11,6 +11,7 @@ use uuid::Uuid;
 
 use crate::agent::session::Session;
 use crate::agent::undo::UndoManager;
+use crate::hooks::HookRegistry;
 
 /// Key for mapping external thread IDs to internal ones.
 #[derive(Clone, Hash, Eq, PartialEq)]
@@ -25,6 +26,7 @@ pub struct SessionManager {
     sessions: RwLock<HashMap<String, Arc<Mutex<Session>>>>,
     thread_map: RwLock<HashMap<ThreadKey, Uuid>>,
     undo_managers: RwLock<HashMap<Uuid, Arc<Mutex<UndoManager>>>>,
+    hooks: Option<Arc<HookRegistry>>,
 }
 
 impl SessionManager {
@@ -34,9 +36,16 @@ impl SessionManager {
             sessions: RwLock::new(HashMap::new()),
             thread_map: RwLock::new(HashMap::new()),
             undo_managers: RwLock::new(HashMap::new()),
+            hooks: None,
         }
     }
 
+    /// Attach a hook registry for session lifecycle events.
+    pub fn with_hooks(mut self, hooks: Arc<HookRegistry>) -> Self {
+        self.hooks = Some(hooks);
+        self
+    }
+
     /// Get or create a session for a user.
     pub async fn get_or_create_session(&self, user_id: &str) -> Arc<Mutex<Session>> {
         // Fast path: check if session exists
@@ -54,8 +63,28 @@ impl SessionManager {
             return Arc::clone(session);
         }
 
-        let session = Arc::new(Mutex::new(Session::new(user_id)));
+        let new_session = Session::new(user_id);
+        let session_id = new_session.id.to_string();
+        let session = Arc::new(Mutex::new(new_session));
         sessions.insert(user_id.to_string(), Arc::clone(&session));
+
+        // Fire OnSessionStart hook (fire-and-forget)
+        if let Some(ref hooks) = self.hooks {
+            let hooks = hooks.clone();
+            let uid = user_id.to_string();
+            let sid = session_id;
+            tokio::spawn(async move {
+                use crate::hooks::HookEvent;
+                let event = HookEvent::SessionStart {
+                    user_id: uid,
+                    session_id: sid,
+                };
+                if let Err(e) = hooks.run(&event).await {
+                    tracing::warn!("OnSessionStart hook error: {}", e);
+                }
+            });
+        }
+
         session
     }
 
@@ -173,8 +202,8 @@ impl SessionManager {
     pub async fn prune_stale_sessions(&self, max_idle: std::time::Duration) -> usize {
         let cutoff = chrono::Utc::now() - chrono::TimeDelta::seconds(max_idle.as_secs() as i64);
 
-        // Find stale session user_ids
-        let stale_users: Vec<String> = {
+        // Find stale sessions (user_id + session_id)
+        let stale_sessions: Vec<(String, String)> = {
             let sessions = self.sessions.read().await;
             sessions
                 .iter()
@@ -182,7 +211,7 @@ impl SessionManager {
                     // Try to lock; skip if contended (someone is actively using it)
                     let sess = session.try_lock().ok()?;
                     if sess.last_active_at < cutoff {
-                        Some(user_id.clone())
+                        Some((user_id.clone(), sess.id.to_string()))
                     } else {
                         None
                     }
@@ -190,6 +219,11 @@ impl SessionManager {
                 .collect()
         };
 
+        let stale_users: Vec<String> = stale_sessions
+            .iter()
+            .map(|(user_id, _)| user_id.clone())
+            .collect();
+
         if stale_users.is_empty() {
             return 0;
         }
@@ -199,14 +233,33 @@ impl SessionManager {
         {
             let sessions = self.sessions.read().await;
             for user_id in &stale_users {
-                if let Some(session) = sessions.get(user_id) {
-                    if let Ok(sess) = session.try_lock() {
-                        stale_thread_ids.extend(sess.threads.keys());
-                    }
+                if let Some(session) = sessions.get(user_id)
+                    && let Ok(sess) = session.try_lock()
+                {
+                    stale_thread_ids.extend(sess.threads.keys());
                 }
             }
         }
 
+        // Fire OnSessionEnd hooks for stale sessions (fire-and-forget)
+        if let Some(ref hooks) = self.hooks {
+            for (user_id, session_id) in &stale_sessions {
+                let hooks = hooks.clone();
+                let uid = user_id.clone();
+                let sid = session_id.clone();
+                tokio::spawn(async move {
+                    use crate::hooks::HookEvent;
+                    let event = HookEvent::SessionEnd {
+                        user_id: uid,
+                        session_id: sid,
+                    };
+                    if let Err(e) = hooks.run(&event).await {
+                        tracing::warn!("OnSessionEnd hook error: {}", e);
+                    }
+                });
+            }
+        }
+
         // Remove sessions
         let count = {
             let mut sessions = self.sessions.write().await;

src/agent/submission.rs

@@ -93,27 +93,26 @@ impl SubmissionParser {
         // /thread <uuid> - switch thread
         if let Some(rest) = lower.strip_prefix("/thread ") {
             let rest = rest.trim();
-            if rest != "new" {
-                if let Ok(id) = Uuid::parse_str(rest) {
-                    return Submission::SwitchThread { thread_id: id };
-                }
+            if rest != "new"
+                && let Ok(id) = Uuid::parse_str(rest)
+            {
+                return Submission::SwitchThread { thread_id: id };
             }
         }
 
         // /resume <uuid> - resume from checkpoint
-        if let Some(rest) = lower.strip_prefix("/resume ") {
-            if let Ok(id) = Uuid::parse_str(rest.trim()) {
-                return Submission::Resume { checkpoint_id: id };
-            }
+        if let Some(rest) = lower.strip_prefix("/resume ")
+            && let Ok(id) = Uuid::parse_str(rest.trim())
+        {
+            return Submission::Resume { checkpoint_id: id };
         }
 
         // Try structured JSON approval (from web gateway's /api/chat/approval endpoint)
-        if trimmed.starts_with('{') {
-            if let Ok(submission) = serde_json::from_str::<Submission>(trimmed) {
-                if matches!(submission, Submission::ExecApproval { .. }) {
-                    return submission;
-                }
-            }
+        if trimmed.starts_with('{')
+            && let Ok(submission) = serde_json::from_str::<Submission>(trimmed)
+            && matches!(submission, Submission::ExecApproval { .. })
+        {
+            return submission;
         }
 
         // Approval responses (simple yes/no/always for pending approvals)

src/agent/undo.rs

@@ -43,6 +43,10 @@ impl Checkpoint {
 }
 
 /// Manager for undo/redo functionality.
+///
+/// Each undo/redo operation pops from one stack and pushes the current state
+/// onto the other, so `undo_count() + redo_count()` stays constant across
+/// undo/redo cycles (only `checkpoint()` and `clear()` change the total).
 pub struct UndoManager {
     /// Stack of past checkpoints (for undo).
     undo_stack: VecDeque<Checkpoint>,
@@ -68,6 +72,14 @@ impl UndoManager {
         self
     }
 
+    /// Push a checkpoint onto the undo stack, trimming oldest entries if over limit.
+    fn push_undo(&mut self, checkpoint: Checkpoint) {
+        self.undo_stack.push_back(checkpoint);
+        while self.undo_stack.len() > self.max_checkpoints {
+            self.undo_stack.pop_front();
+        }
+    }
+
     /// Create a checkpoint at the current state.
     ///
     /// This clears the redo stack since we're creating a new history branch.
@@ -80,24 +92,23 @@ impl UndoManager {
         // Clear redo stack (new branch of history)
         self.redo_stack.clear();
 
-        // Create and push checkpoint
         let checkpoint = Checkpoint::new(turn_number, messages, description);
-        self.undo_stack.push_back(checkpoint);
-
-        // Trim if over limit
-        while self.undo_stack.len() > self.max_checkpoints {
-            self.undo_stack.pop_front();
-        }
+        self.push_undo(checkpoint);
     }
 
     /// Undo: pop the last checkpoint and return it.
     ///
-    /// The current state should be saved to redo stack before calling this.
+    /// Saves the current state to the redo stack and pops the most recent
+    /// checkpoint from the undo stack so that repeated undos walk backwards
+    /// through history.
+    ///
+    /// Takes ownership of `current_messages`; callers must clone first if
+    /// they need to retain a copy.
     pub fn undo(
         &mut self,
         current_turn: usize,
         current_messages: Vec<ChatMessage>,
-    ) -> Option<&Checkpoint> {
+    ) -> Option<Checkpoint> {
         if self.undo_stack.is_empty() {
             return None;
         }
@@ -110,9 +121,8 @@ impl UndoManager {
         );
         self.redo_stack.push(current);
 
-        // Return the most recent checkpoint without removing it
-        // (we keep it so multiple undos can work)
-        self.undo_stack.back()
+        // Pop and return the most recent checkpoint
+        self.undo_stack.pop_back()
     }
 
     /// Pop the last checkpoint from the undo stack.
@@ -121,7 +131,29 @@ impl UndoManager {
     }
 
     /// Redo: restore a previously undone state.
-    pub fn redo(&mut self) -> Option<Checkpoint> {
+    ///
+    /// Saves the current state to the undo stack and pops the most recent
+    /// checkpoint from the redo stack.
+    ///
+    /// Takes ownership of `current_messages`; callers must clone first if
+    /// they need to retain a copy.
+    pub fn redo(
+        &mut self,
+        current_turn: usize,
+        current_messages: Vec<ChatMessage>,
+    ) -> Option<Checkpoint> {
+        if self.redo_stack.is_empty() {
+            return None;
+        }
+
+        // Save current state to undo stack
+        let current = Checkpoint::new(
+            current_turn,
+            current_messages,
+            format!("Turn {}", current_turn),
+        );
+        self.push_undo(current);
+
         self.redo_stack.pop()
     }
 
@@ -214,14 +246,16 @@ mod tests {
         assert!(manager.can_undo());
         assert!(!manager.can_redo());
 
-        // Undo
+        // Undo - returns owned Checkpoint now
         let current = vec![ChatMessage::user("Hello"), ChatMessage::assistant("Hi")];
         let checkpoint = manager.undo(2, current);
         assert!(checkpoint.is_some());
+        let checkpoint = checkpoint.unwrap();
+        assert_eq!(checkpoint.turn_number, 1);
         assert!(manager.can_redo());
 
-        // Redo
-        let restored = manager.redo();
+        // Redo - now requires current state parameters
+        let restored = manager.redo(checkpoint.turn_number, checkpoint.messages);
         assert!(restored.is_some());
     }
 
@@ -249,4 +283,90 @@ mod tests {
         assert!(restored.is_some());
         assert_eq!(manager.undo_count(), 0);
     }
+
+    #[test]
+    fn test_repeated_undo_advances_through_stack() {
+        let mut manager = UndoManager::new();
+
+        // Create 3 checkpoints at turns 0, 1, 2
+        manager.checkpoint(0, vec![], "Turn 0");
+        manager.checkpoint(1, vec![ChatMessage::user("msg1")], "Turn 1");
+        manager.checkpoint(2, vec![ChatMessage::user("msg2")], "Turn 2");
+        assert_eq!(manager.undo_count(), 3);
+
+        // First undo: should return turn 2 checkpoint, stack shrinks to 2
+        let cp1 = manager
+            .undo(3, vec![ChatMessage::user("msg3")])
+            .expect("first undo should succeed");
+        assert_eq!(cp1.turn_number, 2);
+        assert_eq!(manager.undo_count(), 2);
+
+        // Second undo: should return turn 1 checkpoint (different!), stack shrinks to 1
+        let cp2 = manager
+            .undo(cp1.turn_number, cp1.messages)
+            .expect("second undo should succeed");
+        assert_eq!(cp2.turn_number, 1);
+        assert_eq!(manager.undo_count(), 1);
+
+        // Verify we walked backwards through distinct checkpoints
+        assert_ne!(cp1.turn_number, cp2.turn_number);
+    }
+
+    #[test]
+    fn test_undo_redo_cycle_preserves_state() {
+        let mut manager = UndoManager::new();
+
+        let msgs_t0: Vec<ChatMessage> = vec![];
+        let msgs_t1 = vec![ChatMessage::user("hello")];
+        let msgs_t2 = vec![ChatMessage::user("hello"), ChatMessage::assistant("hi")];
+
+        manager.checkpoint(0, msgs_t0, "Turn 0");
+        manager.checkpoint(1, msgs_t1, "Turn 1");
+
+        // Undo from turn 2 -> get turn 1 checkpoint
+        let cp_undo1 = manager
+            .undo(2, msgs_t2.clone())
+            .expect("undo should succeed");
+        assert_eq!(cp_undo1.turn_number, 1);
+
+        // Redo from turn 1 -> get turn 2 state back
+        let cp_redo = manager
+            .redo(cp_undo1.turn_number, cp_undo1.messages)
+            .expect("redo should succeed");
+        assert_eq!(cp_redo.turn_number, 2);
+        assert_eq!(cp_redo.messages.len(), 2);
+
+        // Undo again from turn 2 -> should go back to turn 1 again
+        let cp_undo2 = manager
+            .undo(cp_redo.turn_number, cp_redo.messages)
+            .expect("second undo should succeed");
+        assert_eq!(cp_undo2.turn_number, 1);
+    }
+
+    #[test]
+    fn test_undo_redo_stack_sizes_consistent() {
+        let mut manager = UndoManager::new();
+
+        manager.checkpoint(0, vec![], "Turn 0");
+        manager.checkpoint(1, vec![ChatMessage::user("a")], "Turn 1");
+        manager.checkpoint(2, vec![ChatMessage::user("b")], "Turn 2");
+
+        // Start: undo=3, redo=0, total=3
+        let total = manager.undo_count() + manager.redo_count();
+        assert_eq!(total, 3);
+
+        // After undo: total should still be 3 (one moved from undo to redo,
+        // plus the current state pushed to redo)
+        // Actually: undo pops one (3->2), pushes current to redo (0->1), total=3
+        let cp = manager.undo(3, vec![]).unwrap();
+        assert_eq!(manager.undo_count() + manager.redo_count(), 3);
+
+        // After redo: redo pops one (1->0), pushes current to undo (2->3), total=3
+        let cp2 = manager.redo(cp.turn_number, cp.messages).unwrap();
+        assert_eq!(manager.undo_count() + manager.redo_count(), 3);
+
+        // After another undo: same invariant
+        let _cp3 = manager.undo(cp2.turn_number, cp2.messages).unwrap();
+        assert_eq!(manager.undo_count() + manager.redo_count(), 3);
+    }
 }

src/agent/worker.rs

@@ -10,8 +10,9 @@ use uuid::Uuid;
 use crate::agent::scheduler::WorkerMessage;
 use crate::agent::task::TaskOutput;
 use crate::context::{ContextManager, JobState};
+use crate::db::Database;
 use crate::error::Error;
-use crate::history::Store;
+use crate::hooks::HookRegistry;
 use crate::llm::{
     ActionPlan, ChatMessage, LlmProvider, Reasoning, ReasoningContext, RespondResult, ToolSelection,
 };
@@ -28,7 +29,8 @@ pub struct WorkerDeps {
     pub llm: Arc<dyn LlmProvider>,
     pub safety: Arc<SafetyLayer>,
     pub tools: Arc<ToolRegistry>,
-    pub store: Option<Arc<Store>>,
+    pub store: Option<Arc<dyn Database>>,
+    pub hooks: Arc<HookRegistry>,
     pub timeout: Duration,
     pub use_planning: bool,
 }
@@ -67,7 +69,7 @@ impl Worker {
         &self.deps.tools
     }
 
-    fn store(&self) -> Option<&Arc<Store>> {
+    fn store(&self) -> Option<&Arc<dyn Database>> {
         self.deps.store.as_ref()
     }
 
@@ -227,11 +229,11 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
             }
 
             // Check for cancellation
-            if let Ok(ctx) = self.context_manager().get_context(self.job_id).await {
-                if ctx.state == JobState::Cancelled {
-                    tracing::info!("Worker for job {} detected cancellation", self.job_id);
-                    return Ok(());
-                }
+            if let Ok(ctx) = self.context_manager().get_context(self.job_id).await
+                && ctx.state == JobState::Cancelled
+            {
+                tracing::info!("Worker for job {} detected cancellation", self.job_id);
+                return Ok(());
             }
 
             iteration += 1;
@@ -248,16 +250,15 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
 
             if selections.is_empty() {
                 // No tools from select_tools, ask LLM directly (may still return tool calls)
-                let respond_result = reasoning.respond_with_tools(reason_ctx).await?;
+                let respond_output = reasoning.respond_with_tools(reason_ctx).await?;
 
-                match respond_result {
+                match respond_output.result {
                     RespondResult::Text(response) => {
-                        // Check for completion keywords
-                        let response_lower = response.to_lowercase();
-                        if response_lower.contains("complete")
-                            || response_lower.contains("finished")
-                            || response_lower.contains("done")
-                        {
+                        // Check for explicit completion phrases. Use word-boundary
+                        // aware checks to avoid false positives like "incomplete",
+                        // "not done", or "unfinished". Only the LLM's own response
+                        // (not tool output) can trigger this.
+                        if crate::util::llm_signals_completion(&response) {
                             self.mark_completed().await?;
                             return Ok(());
                         }
@@ -300,6 +301,7 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
                                 parameters: tc.arguments.clone(),
                                 reasoning: String::new(),
                                 alternatives: vec![],
+                                tool_call_id: tc.id.clone(),
                             };
 
                             self.process_tool_result(reason_ctx, &selection, result)
@@ -352,23 +354,11 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
             .map(|selection| {
                 let tool_name = selection.tool_name.clone();
                 let params = selection.parameters.clone();
-                let tools = self.tools().clone();
-                let context_manager = self.context_manager().clone();
-                let safety = self.safety().clone();
+                let deps = self.deps.clone();
                 let job_id = self.job_id;
-                let store = self.deps.store.clone();
 
                 async move {
-                    let result = Self::execute_tool_inner(
-                        tools,
-                        context_manager,
-                        safety,
-                        store,
-                        job_id,
-                        &tool_name,
-                        &params,
-                    )
-                    .await;
+                    let result = Self::execute_tool_inner(&deps, job_id, &tool_name, &params).await;
                     ToolExecResult { result }
                 }
             })
@@ -379,20 +369,18 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
 
     /// Inner tool execution logic that can be called from both single and parallel paths.
     async fn execute_tool_inner(
-        tools: Arc<ToolRegistry>,
-        context_manager: Arc<ContextManager>,
-        safety: Arc<SafetyLayer>,
-        store: Option<Arc<Store>>,
+        deps: &WorkerDeps,
         job_id: Uuid,
         tool_name: &str,
         params: &serde_json::Value,
     ) -> Result<String, Error> {
-        let tool = tools
-            .get(tool_name)
-            .await
-            .ok_or_else(|| crate::error::ToolError::NotFound {
-                name: tool_name.to_string(),
-            })?;
+        let tool =
+            deps.tools
+                .get(tool_name)
+                .await
+                .ok_or_else(|| crate::error::ToolError::NotFound {
+                    name: tool_name.to_string(),
+                })?;
 
         // Tools requiring approval are blocked in autonomous jobs
         if tool.requires_approval() {
@@ -402,8 +390,46 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
             .into());
         }
 
-        // Get job context for the tool
-        let job_ctx = context_manager.get_context(job_id).await?;
+        // Fetch job context early so we have the real user_id for hooks
+        let job_ctx = deps.context_manager.get_context(job_id).await?;
+
+        // Run BeforeToolCall hook
+        let params = {
+            use crate::hooks::{HookError, HookEvent, HookOutcome};
+            let event = HookEvent::ToolCall {
+                tool_name: tool_name.to_string(),
+                parameters: params.clone(),
+                user_id: job_ctx.user_id.clone(),
+                context: format!("job:{}", job_id),
+            };
+            match deps.hooks.run(&event).await {
+                Err(HookError::Rejected { reason }) => {
+                    return Err(crate::error::ToolError::ExecutionFailed {
+                        name: tool_name.to_string(),
+                        reason: format!("Blocked by hook: {}", reason),
+                    }
+                    .into());
+                }
+                Err(err) => {
+                    return Err(crate::error::ToolError::ExecutionFailed {
+                        name: tool_name.to_string(),
+                        reason: format!("Blocked by hook failure mode: {}", err),
+                    }
+                    .into());
+                }
+                Ok(HookOutcome::Continue {
+                    modified: Some(new_params),
+                }) => serde_json::from_str(&new_params).unwrap_or_else(|e| {
+                    tracing::warn!(
+                        tool = %tool_name,
+                        "Hook returned non-JSON modification for ToolCall, ignoring: {}",
+                        e
+                    );
+                    params.clone()
+                }),
+                _ => params.clone(),
+            }
+        };
         if job_ctx.state == JobState::Cancelled {
             return Err(crate::error::ToolError::ExecutionFailed {
                 name: tool_name.to_string(),
@@ -413,7 +439,7 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
         }
 
         // Validate tool parameters
-        let validation = safety.validator().validate_tool_params(params);
+        let validation = deps.safety.validator().validate_tool_params(&params);
         if !validation.is_valid {
             let details = validation
                 .errors
@@ -478,8 +504,8 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
             Ok(Ok(output)) => {
                 let output_str = serde_json::to_string_pretty(&output.result)
                     .ok()
-                    .map(|s| safety.sanitize_tool_output(tool_name, &s).content);
-                context_manager
+                    .map(|s| deps.safety.sanitize_tool_output(tool_name, &s).content);
+                deps.context_manager
                     .update_memory(job_id, |mem| {
                         let rec = mem.create_action(tool_name, params.clone()).succeed(
                             output_str.clone(),
@@ -492,7 +518,8 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
                     .await
                     .ok()
             }
-            Ok(Err(e)) => context_manager
+            Ok(Err(e)) => deps
+                .context_manager
                 .update_memory(job_id, |mem| {
                     let rec = mem
                         .create_action(tool_name, params.clone())
@@ -502,7 +529,8 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
                 })
                 .await
                 .ok(),
-            Err(_) => context_manager
+            Err(_) => deps
+                .context_manager
                 .update_memory(job_id, |mem| {
                     let rec = mem
                         .create_action(tool_name, params.clone())
@@ -515,7 +543,7 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
         };
 
         // Persist action to database (fire-and-forget)
-        if let (Some(action), Some(store)) = (action, store) {
+        if let (Some(action), Some(store)) = (action, deps.store.clone()) {
             tokio::spawn(async move {
                 if let Err(e) = store.save_action(job_id, &action).await {
                     tracing::warn!("Failed to persist action for job {}: {}", job_id, e);
@@ -566,17 +594,14 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
                 );
 
                 reason_ctx.messages.push(ChatMessage::tool_result(
-                    "tool_call_id",
+                    &selection.tool_call_id,
                     &selection.tool_name,
                     wrapped,
                 ));
 
-                // Check if job is complete
-                if output.contains("TASK_COMPLETE") || output.contains("JOB_DONE") {
-                    self.mark_completed().await?;
-                    return Ok(true);
-                }
-
+                // Tool output never drives job completion. A malicious tool could
+                // emit "TASK_COMPLETE" to force premature completion. Only the LLM's
+                // own structured response (in execution_loop) can mark a job done.
                 Ok(false)
             }
             Err(e) => {
@@ -601,7 +626,7 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
                 }
 
                 reason_ctx.messages.push(ChatMessage::tool_result(
-                    "tool_call_id",
+                    &selection.tool_call_id,
                     &selection.tool_name,
                     format!("Error: {}", e),
                 ));
@@ -651,12 +676,15 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
                 .execute_tool(&action.tool_name, &action.parameters)
                 .await;
 
-            // Create a synthetic ToolSelection for process_tool_result
+            // Create a synthetic ToolSelection for process_tool_result.
+            // Plan actions don't originate from an LLM tool_call response so
+            // there is no real tool_call_id; generate a unique one.
             let selection = ToolSelection {
                 tool_name: action.tool_name.clone(),
                 parameters: action.parameters.clone(),
                 reasoning: action.reasoning.clone(),
                 alternatives: vec![],
+                tool_call_id: format!("plan_{}_{}", self.job_id, i),
             };
 
             // Process the result
@@ -680,11 +708,7 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
         let response = reasoning.respond(reason_ctx).await?;
         reason_ctx.messages.push(ChatMessage::assistant(&response));
 
-        let response_lower = response.to_lowercase();
-        if response_lower.contains("complete")
-            || response_lower.contains("finished")
-            || response_lower.contains("done")
-        {
+        if crate::util::llm_signals_completion(&response) {
             self.mark_completed().await?;
         } else {
             // Job not complete, could re-plan or fall back to direct selection
@@ -705,16 +729,7 @@ Report when the job is complete or if you encounter issues you cannot resolve."#
         tool_name: &str,
         params: &serde_json::Value,
     ) -> Result<String, Error> {
-        Self::execute_tool_inner(
-            self.tools().clone(),
-            self.context_manager().clone(),
-            self.safety().clone(),
-            self.deps.store.clone(),
-            self.job_id,
-            tool_name,
-            params,
-        )
-        .await
+        Self::execute_tool_inner(&self.deps, self.job_id, tool_name, params).await
     }
 
     async fn mark_completed(&self) -> Result<(), Error> {
@@ -779,3 +794,86 @@ impl From<TaskOutput> for Result<String, Error> {
         })
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use crate::llm::ToolSelection;
+    use crate::util::llm_signals_completion;
+
+    #[test]
+    fn test_tool_selection_preserves_call_id() {
+        let selection = ToolSelection {
+            tool_name: "memory_search".to_string(),
+            parameters: serde_json::json!({"query": "test"}),
+            reasoning: "Need to search memory".to_string(),
+            alternatives: vec![],
+            tool_call_id: "call_abc123".to_string(),
+        };
+
+        assert_eq!(selection.tool_call_id, "call_abc123");
+        assert_ne!(
+            selection.tool_call_id, "tool_call_id",
+            "tool_call_id must not be the hardcoded placeholder string"
+        );
+    }
+
+    #[test]
+    fn test_completion_positive_signals() {
+        assert!(llm_signals_completion("The job is complete."));
+        assert!(llm_signals_completion(
+            "I have completed the task successfully."
+        ));
+        assert!(llm_signals_completion("The task is done."));
+        assert!(llm_signals_completion("The task is finished."));
+        assert!(llm_signals_completion(
+            "All steps are complete and verified."
+        ));
+        assert!(llm_signals_completion(
+            "I've done all the work. The work is done."
+        ));
+        assert!(llm_signals_completion(
+            "Successfully completed the migration."
+        ));
+    }
+
+    #[test]
+    fn test_completion_negative_signals_block_false_positives() {
+        // These contain completion keywords but also negation, should NOT trigger.
+        assert!(!llm_signals_completion("The task is not complete yet."));
+        assert!(!llm_signals_completion("This is not done."));
+        assert!(!llm_signals_completion("The work is incomplete."));
+        assert!(!llm_signals_completion(
+            "The migration is not yet finished."
+        ));
+        assert!(!llm_signals_completion("The job isn't done yet."));
+        assert!(!llm_signals_completion("This remains unfinished."));
+    }
+
+    #[test]
+    fn test_completion_does_not_match_bare_substrings() {
+        // Bare words embedded in other text should NOT trigger completion.
+        assert!(!llm_signals_completion(
+            "I need to complete more work first."
+        ));
+        assert!(!llm_signals_completion(
+            "Let me finish the remaining steps."
+        ));
+        assert!(!llm_signals_completion(
+            "I'm done analyzing, now let me fix it."
+        ));
+        assert!(!llm_signals_completion(
+            "I completed step 1 but step 2 remains."
+        ));
+    }
+
+    #[test]
+    fn test_completion_tool_output_injection() {
+        // A malicious tool output echoed by the LLM should not trigger
+        // completion unless it forms a genuine completion phrase.
+        assert!(!llm_signals_completion("TASK_COMPLETE"));
+        assert!(!llm_signals_completion("JOB_DONE"));
+        assert!(!llm_signals_completion(
+            "The tool returned: TASK_COMPLETE signal"
+        ));
+    }
+}

src/boot_screen.rs

@@ -0,0 +1,208 @@
+//! Boot screen displayed after all initialization completes.
+//!
+//! Shows a polished ANSI-styled status panel summarizing the agent's runtime
+//! state: model, database, tool count, enabled features, active channels,
+//! and the gateway URL.
+
+/// All displayable fields for the boot screen.
+pub struct BootInfo {
+    pub version: String,
+    pub agent_name: String,
+    pub llm_backend: String,
+    pub llm_model: String,
+    pub cheap_model: Option<String>,
+    pub db_backend: String,
+    pub db_connected: bool,
+    pub tool_count: usize,
+    pub gateway_url: Option<String>,
+    pub embeddings_enabled: bool,
+    pub embeddings_provider: Option<String>,
+    pub heartbeat_enabled: bool,
+    pub heartbeat_interval_secs: u64,
+    pub sandbox_enabled: bool,
+    pub claude_code_enabled: bool,
+    pub routines_enabled: bool,
+    pub channels: Vec<String>,
+}
+
+/// Print the boot screen to stdout.
+pub fn print_boot_screen(info: &BootInfo) {
+    // ANSI codes matching existing REPL palette
+    let bold = "\x1b[1m";
+    let cyan = "\x1b[36m";
+    let dim = "\x1b[90m";
+    let yellow_underline = "\x1b[33;4m";
+    let reset = "\x1b[0m";
+
+    let border = format!("  {dim}{}{reset}", "\u{2576}".repeat(58));
+
+    println!();
+    println!("{border}");
+    println!();
+    println!("  {bold}{}{reset} v{}", info.agent_name, info.version);
+    println!();
+
+    // Model line
+    let model_display = if let Some(ref cheap) = info.cheap_model {
+        format!(
+            "{cyan}{}{reset}  {dim}cheap{reset} {cyan}{}{reset}",
+            info.llm_model, cheap
+        )
+    } else {
+        format!("{cyan}{}{reset}", info.llm_model)
+    };
+    println!(
+        "  {dim}model{reset}     {model_display}  {dim}via {}{reset}",
+        info.llm_backend
+    );
+
+    // Database line
+    let db_status = if info.db_connected {
+        "connected"
+    } else {
+        "none"
+    };
+    println!(
+        "  {dim}database{reset}  {cyan}{}{reset} {dim}({db_status}){reset}",
+        info.db_backend
+    );
+
+    // Tools line
+    println!(
+        "  {dim}tools{reset}     {cyan}{}{reset} {dim}registered{reset}",
+        info.tool_count
+    );
+
+    // Features line
+    let mut features = Vec::new();
+    if info.embeddings_enabled {
+        if let Some(ref provider) = info.embeddings_provider {
+            features.push(format!("embeddings ({provider})"));
+        } else {
+            features.push("embeddings".to_string());
+        }
+    }
+    if info.heartbeat_enabled {
+        let mins = info.heartbeat_interval_secs / 60;
+        features.push(format!("heartbeat ({mins}m)"));
+    }
+    if info.sandbox_enabled {
+        features.push("sandbox".to_string());
+    }
+    if info.claude_code_enabled {
+        features.push("claude-code".to_string());
+    }
+    if info.routines_enabled {
+        features.push("routines".to_string());
+    }
+    if !features.is_empty() {
+        println!(
+            "  {dim}features{reset}  {cyan}{}{reset}",
+            features.join("  ")
+        );
+    }
+
+    // Channels line
+    if !info.channels.is_empty() {
+        println!(
+            "  {dim}channels{reset}  {cyan}{}{reset}",
+            info.channels.join("  ")
+        );
+    }
+
+    // Gateway URL (highlighted)
+    if let Some(ref url) = info.gateway_url {
+        println!();
+        println!("  {dim}gateway{reset}   {yellow_underline}{url}{reset}");
+    }
+
+    println!();
+    println!("{border}");
+    println!();
+    println!("  /help for commands, /quit to exit");
+    println!();
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_print_boot_screen_full() {
+        let info = BootInfo {
+            version: "0.2.0".to_string(),
+            agent_name: "ironclaw".to_string(),
+            llm_backend: "nearai".to_string(),
+            llm_model: "claude-3-5-sonnet-20241022".to_string(),
+            cheap_model: Some("gpt-4o-mini".to_string()),
+            db_backend: "libsql".to_string(),
+            db_connected: true,
+            tool_count: 24,
+            gateway_url: Some("http://127.0.0.1:3001/?token=abc123".to_string()),
+            embeddings_enabled: true,
+            embeddings_provider: Some("openai".to_string()),
+            heartbeat_enabled: true,
+            heartbeat_interval_secs: 1800,
+            sandbox_enabled: true,
+            claude_code_enabled: false,
+            routines_enabled: true,
+            channels: vec![
+                "repl".to_string(),
+                "gateway".to_string(),
+                "telegram".to_string(),
+            ],
+        };
+        // Should not panic
+        print_boot_screen(&info);
+    }
+
+    #[test]
+    fn test_print_boot_screen_minimal() {
+        let info = BootInfo {
+            version: "0.2.0".to_string(),
+            agent_name: "ironclaw".to_string(),
+            llm_backend: "nearai".to_string(),
+            llm_model: "gpt-4o".to_string(),
+            cheap_model: None,
+            db_backend: "none".to_string(),
+            db_connected: false,
+            tool_count: 5,
+            gateway_url: None,
+            embeddings_enabled: false,
+            embeddings_provider: None,
+            heartbeat_enabled: false,
+            heartbeat_interval_secs: 0,
+            sandbox_enabled: false,
+            claude_code_enabled: false,
+            routines_enabled: false,
+            channels: vec![],
+        };
+        // Should not panic
+        print_boot_screen(&info);
+    }
+
+    #[test]
+    fn test_print_boot_screen_no_features() {
+        let info = BootInfo {
+            version: "0.1.0".to_string(),
+            agent_name: "test".to_string(),
+            llm_backend: "openai".to_string(),
+            llm_model: "gpt-4o".to_string(),
+            cheap_model: None,
+            db_backend: "postgres".to_string(),
+            db_connected: true,
+            tool_count: 10,
+            gateway_url: None,
+            embeddings_enabled: false,
+            embeddings_provider: None,
+            heartbeat_enabled: false,
+            heartbeat_interval_secs: 0,
+            sandbox_enabled: false,
+            claude_code_enabled: false,
+            routines_enabled: false,
+            channels: vec!["repl".to_string()],
+        };
+        // Should not panic
+        print_boot_screen(&info);
+    }
+}

src/bootstrap.rs

@@ -1,147 +1,145 @@
-//! Bootstrap configuration for IronClaw.
+//! Bootstrap helpers for IronClaw.
 //!
-//! These are the only settings that MUST live on disk because they're needed
-//! before the database connection is established. Everything else lives in the
-//! `settings` table in PostgreSQL.
+//! The only setting that truly needs disk persistence before the database is
+//! available is `DATABASE_URL` (chicken-and-egg: can't connect to DB without
+//! it). Everything else is auto-detected or read from env vars.
 //!
-//! File: `~/.ironclaw/bootstrap.json`
+//! File: `~/.ironclaw/.env` (standard dotenvy format)
 
 use std::path::PathBuf;
 
-use serde::{Deserialize, Serialize};
-
-use crate::settings::KeySource;
-
-/// Minimal config needed to connect to the database and decrypt secrets.
-///
-/// This is the only JSON file IronClaw reads from disk at startup.
-/// All other configuration lives in the `settings` table in PostgreSQL.
-#[derive(Debug, Clone, Serialize, Deserialize)]
-pub struct BootstrapConfig {
-    /// Database connection URL (postgres://...).
-    #[serde(default)]
-    pub database_url: Option<String>,
-
-    /// Database connection pool size.
-    #[serde(default)]
-    pub database_pool_size: Option<usize>,
-
-    /// Source for the secrets master key.
-    #[serde(default)]
-    pub secrets_master_key_source: KeySource,
-
-    /// Whether onboarding wizard has been completed.
-    #[serde(default)]
-    pub onboard_completed: bool,
+/// Path to the IronClaw-specific `.env` file: `~/.ironclaw/.env`.
+pub fn ironclaw_env_path() -> PathBuf {
+    dirs::home_dir()
+        .unwrap_or_else(|| PathBuf::from("."))
+        .join(".ironclaw")
+        .join(".env")
 }
 
-impl Default for BootstrapConfig {
-    fn default() -> Self {
-        Self {
-            database_url: None,
-            database_pool_size: None,
-            secrets_master_key_source: KeySource::None,
-            onboard_completed: false,
-        }
+/// Load env vars from `~/.ironclaw/.env` (in addition to the standard `.env`).
+///
+/// Call this **after** `dotenvy::dotenv()` so that the standard `./.env`
+/// takes priority over `~/.ironclaw/.env`. dotenvy never overwrites
+/// existing env vars, so the effective priority is:
+///
+///   explicit env vars > `./.env` > `~/.ironclaw/.env`
+///
+/// If `~/.ironclaw/.env` doesn't exist but the legacy `bootstrap.json` does,
+/// extracts `DATABASE_URL` from it and writes the `.env` file (one-time
+/// upgrade from the old config format).
+pub fn load_ironclaw_env() {
+    let path = ironclaw_env_path();
+
+    if !path.exists() {
+        // One-time upgrade: extract DATABASE_URL from legacy bootstrap.json
+        migrate_bootstrap_json_to_env(&path);
+    }
+
+    if path.exists() {
+        let _ = dotenvy::from_path(&path);
     }
 }
 
-impl BootstrapConfig {
-    /// Default bootstrap file path: `~/.ironclaw/bootstrap.json`.
-    pub fn default_path() -> PathBuf {
-        dirs::home_dir()
-            .unwrap_or_else(|| PathBuf::from("."))
-            .join(".ironclaw")
-            .join("bootstrap.json")
+/// If `bootstrap.json` exists, pull `database_url` out of it and write `.env`.
+fn migrate_bootstrap_json_to_env(env_path: &std::path::Path) {
+    let ironclaw_dir = env_path
+        .parent()
+        .unwrap_or_else(|| std::path::Path::new("."));
+    let bootstrap_path = ironclaw_dir.join("bootstrap.json");
+
+    if !bootstrap_path.exists() {
+        return;
     }
 
-    /// Legacy settings.json path (for migration detection).
-    pub fn legacy_settings_path() -> PathBuf {
-        dirs::home_dir()
-            .unwrap_or_else(|| PathBuf::from("."))
-            .join(".ironclaw")
-            .join("settings.json")
-    }
+    let content = match std::fs::read_to_string(&bootstrap_path) {
+        Ok(c) => c,
+        Err(_) => return,
+    };
 
-    /// Load from the default path, falling back to legacy settings.json,
-    /// then to defaults if neither exists.
-    pub fn load() -> Self {
-        let bootstrap_path = Self::default_path();
-        if bootstrap_path.exists() {
-            return Self::load_from(&bootstrap_path);
+    // Minimal parse: just grab database_url from the JSON
+    let parsed: serde_json::Value = match serde_json::from_str(&content) {
+        Ok(v) => v,
+        Err(_) => return,
+    };
+
+    if let Some(url) = parsed.get("database_url").and_then(|v| v.as_str()) {
+        if let Some(parent) = env_path.parent()
+            && let Err(e) = std::fs::create_dir_all(parent)
+        {
+            eprintln!("Warning: failed to create {}: {}", parent.display(), e);
+            return;
         }
-
-        // Fall back to legacy settings.json (extract just the 4 bootstrap fields)
-        let legacy_path = Self::legacy_settings_path();
-        if legacy_path.exists() {
-            return Self::load_from_legacy(&legacy_path);
+        if let Err(e) = std::fs::write(env_path, format!("DATABASE_URL=\"{}\"\n", url)) {
+            eprintln!("Warning: failed to migrate bootstrap.json to .env: {}", e);
+            return;
         }
-
-        Self::default()
-    }
-
-    /// Load from a specific path.
-    pub fn load_from(path: &PathBuf) -> Self {
-        match std::fs::read_to_string(path) {
-            Ok(data) => serde_json::from_str(&data).unwrap_or_default(),
-            Err(_) => Self::default(),
-        }
-    }
-
-    /// Extract bootstrap fields from a legacy settings.json.
-    fn load_from_legacy(path: &PathBuf) -> Self {
-        match std::fs::read_to_string(path) {
-            Ok(data) => {
-                // The legacy Settings struct is a superset; serde will ignore extra fields.
-                serde_json::from_str(&data).unwrap_or_default()
-            }
-            Err(_) => Self::default(),
-        }
-    }
-
-    /// Save to the default path.
-    pub fn save(&self) -> std::io::Result<()> {
-        self.save_to(&Self::default_path())
-    }
-
-    /// Save to a specific path.
-    pub fn save_to(&self, path: &PathBuf) -> std::io::Result<()> {
-        if let Some(parent) = path.parent() {
-            std::fs::create_dir_all(parent)?;
-        }
-        let json = serde_json::to_string_pretty(self)
-            .map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?;
-        std::fs::write(path, json)
+        rename_to_migrated(&bootstrap_path);
+        eprintln!(
+            "Migrated DATABASE_URL from bootstrap.json to {}",
+            env_path.display()
+        );
     }
 }
 
-/// One-time migration from disk config files to the database settings table.
+/// Write database bootstrap vars to `~/.ironclaw/.env`.
 ///
-/// On first boot after upgrade, checks if:
-/// 1. `~/.ironclaw/settings.json` exists
-/// 2. The DB settings table is empty for this user
+/// These settings form the chicken-and-egg layer: they must be available
+/// from the filesystem (env vars) BEFORE any database connection, because
+/// they determine which database to connect to. Everything else is stored
+/// in the database itself.
 ///
-/// If both conditions hold, migrates settings, MCP servers, and session data
-/// to the database, writes `bootstrap.json`, and renames old files to `.migrated`.
+/// Creates the parent directory if it doesn't exist.
+/// Values are double-quoted so that `#` (common in URL-encoded passwords)
+/// and other shell-special characters are preserved by dotenvy.
+pub fn save_bootstrap_env(vars: &[(&str, &str)]) -> std::io::Result<()> {
+    let path = ironclaw_env_path();
+    if let Some(parent) = path.parent() {
+        std::fs::create_dir_all(parent)?;
+    }
+    let mut content = String::new();
+    for (key, value) in vars {
+        content.push_str(&format!("{}=\"{}\"\n", key, value));
+    }
+    std::fs::write(&path, content)
+}
+
+/// Write `DATABASE_URL` to `~/.ironclaw/.env`.
+///
+/// Convenience wrapper around `save_bootstrap_env` for single-value migration
+/// paths. Prefer `save_bootstrap_env` for new code.
+pub fn save_database_url(url: &str) -> std::io::Result<()> {
+    save_bootstrap_env(&[("DATABASE_URL", url)])
+}
+
+/// One-time migration of legacy `~/.ironclaw/settings.json` into the database.
+///
+/// Only runs when a `settings.json` exists on disk AND the DB has no settings
+/// yet. After the wizard writes directly to the DB, this path is only hit by
+/// users upgrading from the old disk-only configuration.
+///
+/// After syncing, renames `settings.json` to `.migrated` so it won't trigger again.
 pub async fn migrate_disk_to_db(
-    store: &crate::history::Store,
+    store: &dyn crate::db::Database,
     user_id: &str,
 ) -> Result<(), MigrationError> {
-    let legacy_settings_path = BootstrapConfig::legacy_settings_path();
+    let ironclaw_dir = dirs::home_dir()
+        .unwrap_or_else(|| PathBuf::from("."))
+        .join(".ironclaw");
+    let legacy_settings_path = ironclaw_dir.join("settings.json");
+
     if !legacy_settings_path.exists() {
         tracing::debug!("No legacy settings.json found, skipping disk-to-DB migration");
         return Ok(());
     }
 
-    // Only migrate if DB is empty for this user
+    // If DB already has settings, this is not a first boot, the wizard already
+    // wrote directly to the DB. Just clean up the stale file.
     let has_settings = store.has_settings(user_id).await.map_err(|e| {
         MigrationError::Database(format!("Failed to check existing settings: {}", e))
     })?;
     if has_settings {
-        tracing::debug!(
-            "DB already has settings for user '{}', skipping migration",
-            user_id
-        );
+        tracing::info!("DB already has settings, renaming stale settings.json");
+        rename_to_migrated(&legacy_settings_path);
         return Ok(());
     }
 
@@ -160,22 +158,14 @@ pub async fn migrate_disk_to_db(
         tracing::info!("Migrated {} settings to database", db_map.len());
     }
 
-    // 2. Write bootstrap.json with the 4 essential fields
-    let bootstrap = BootstrapConfig {
-        database_url: settings.database_url.clone(),
-        database_pool_size: settings.database_pool_size,
-        secrets_master_key_source: settings.secrets_master_key_source,
-        onboard_completed: settings.onboard_completed,
-    };
-    bootstrap
-        .save()
-        .map_err(|e| MigrationError::Io(format!("Failed to write bootstrap.json: {}", e)))?;
-    tracing::info!("Wrote bootstrap.json");
+    // 2. Write DATABASE_URL to ~/.ironclaw/.env
+    if let Some(ref url) = settings.database_url {
+        save_database_url(url)
+            .map_err(|e| MigrationError::Io(format!("Failed to write .env: {}", e)))?;
+        tracing::info!("Wrote DATABASE_URL to {}", ironclaw_env_path().display());
+    }
 
     // 3. Migrate mcp-servers.json if it exists
-    let ironclaw_dir = dirs::home_dir()
-        .unwrap_or_else(|| PathBuf::from("."))
-        .join(".ironclaw");
     let mcp_path = ironclaw_dir.join("mcp-servers.json");
     if mcp_path.exists() {
         match std::fs::read_to_string(&mcp_path) {
@@ -211,7 +201,7 @@ pub async fn migrate_disk_to_db(
             Ok(content) => match serde_json::from_str::<serde_json::Value>(&content) {
                 Ok(value) => {
                     store
-                        .set_setting(user_id, "nearai.session", &value)
+                        .set_setting(user_id, "nearai.session_token", &value)
                         .await
                         .map_err(|e| {
                             MigrationError::Database(format!(
@@ -236,12 +226,19 @@ pub async fn migrate_disk_to_db(
     // 5. Rename settings.json to .migrated (don't delete, safety net)
     rename_to_migrated(&legacy_settings_path);
 
+    // 6. Clean up old bootstrap.json if it exists (superseded by .env)
+    let old_bootstrap = ironclaw_dir.join("bootstrap.json");
+    if old_bootstrap.exists() {
+        rename_to_migrated(&old_bootstrap);
+        tracing::info!("Renamed old bootstrap.json to .migrated");
+    }
+
     tracing::info!("Disk-to-DB migration complete");
     Ok(())
 }
 
 /// Rename a file to `<name>.migrated` as a safety net.
-fn rename_to_migrated(path: &PathBuf) {
+fn rename_to_migrated(path: &std::path::Path) {
     let mut migrated = path.as_os_str().to_owned();
     migrated.push(".migrated");
     if let Err(e) = std::fs::rename(path, &migrated) {
@@ -264,62 +261,204 @@ mod tests {
     use tempfile::tempdir;
 
     #[test]
-    fn test_bootstrap_save_load() {
+    fn test_save_and_load_database_url() {
         let dir = tempdir().unwrap();
-        let path = dir.path().join("bootstrap.json");
+        let env_path = dir.path().join(".env");
 
-        let config = BootstrapConfig {
-            database_url: Some("postgres://localhost/test".to_string()),
-            database_pool_size: Some(5),
-            secrets_master_key_source: KeySource::Keychain,
-            onboard_completed: true,
-        };
+        // Write in the quoted format that save_database_url uses
+        let url = "postgres://localhost:5432/ironclaw_test";
+        std::fs::write(&env_path, format!("DATABASE_URL=\"{}\"\n", url)).unwrap();
 
-        config.save_to(&path).unwrap();
-
-        let loaded = BootstrapConfig::load_from(&path);
+        // Verify the content is a valid dotenv line (quoted)
+        let content = std::fs::read_to_string(&env_path).unwrap();
         assert_eq!(
-            loaded.database_url,
-            Some("postgres://localhost/test".to_string())
+            content,
+            "DATABASE_URL=\"postgres://localhost:5432/ironclaw_test\"\n"
         );
-        assert_eq!(loaded.database_pool_size, Some(5));
-        assert_eq!(loaded.secrets_master_key_source, KeySource::Keychain);
-        assert!(loaded.onboard_completed);
+
+        // Verify dotenvy can parse it (strips quotes automatically)
+        let parsed: Vec<(String, String)> = dotenvy::from_path_iter(&env_path)
+            .unwrap()
+            .filter_map(|r| r.ok())
+            .collect();
+        assert_eq!(parsed.len(), 1);
+        assert_eq!(parsed[0].0, "DATABASE_URL");
+        assert_eq!(parsed[0].1, url);
     }
 
     #[test]
-    fn test_bootstrap_from_legacy_settings() {
+    fn test_save_database_url_with_hash_in_password() {
         let dir = tempdir().unwrap();
-        let path = dir.path().join("settings.json");
+        let env_path = dir.path().join(".env");
 
-        // Write a legacy settings.json with many extra fields
-        let legacy = serde_json::json!({
-            "database_url": "postgres://localhost/ironclaw",
-            "database_pool_size": 10,
+        // URLs with # in the password are common (URL-encoded special chars).
+        // Without quoting, dotenvy treats # as a comment delimiter.
+        let url = "postgres://user:p%23ss@localhost:5432/ironclaw";
+        std::fs::write(&env_path, format!("DATABASE_URL=\"{}\"\n", url)).unwrap();
+
+        let parsed: Vec<(String, String)> = dotenvy::from_path_iter(&env_path)
+            .unwrap()
+            .filter_map(|r| r.ok())
+            .collect();
+        assert_eq!(parsed.len(), 1);
+        assert_eq!(parsed[0].0, "DATABASE_URL");
+        assert_eq!(parsed[0].1, url);
+    }
+
+    #[test]
+    fn test_save_database_url_creates_parent_dirs() {
+        let dir = tempdir().unwrap();
+        let nested = dir.path().join("deep").join("nested");
+        let env_path = nested.join(".env");
+
+        // Parent doesn't exist yet
+        assert!(!nested.exists());
+
+        // The global function uses a fixed path, so we test the logic directly
+        std::fs::create_dir_all(&nested).unwrap();
+        std::fs::write(&env_path, "DATABASE_URL=postgres://test\n").unwrap();
+
+        assert!(env_path.exists());
+        let content = std::fs::read_to_string(&env_path).unwrap();
+        assert!(content.contains("DATABASE_URL=postgres://test"));
+    }
+
+    #[test]
+    fn test_ironclaw_env_path() {
+        let path = ironclaw_env_path();
+        assert!(path.ends_with(".ironclaw/.env"));
+    }
+
+    #[test]
+    fn test_migrate_bootstrap_json_to_env() {
+        let dir = tempdir().unwrap();
+        let env_path = dir.path().join(".env");
+        let bootstrap_path = dir.path().join("bootstrap.json");
+
+        // Write a legacy bootstrap.json
+        let bootstrap_json = serde_json::json!({
+            "database_url": "postgres://localhost/ironclaw_upgrade",
+            "database_pool_size": 5,
             "secrets_master_key_source": "keychain",
-            "onboard_completed": true,
-            "selected_model": "claude-3-5-sonnet",
-            "agent": { "name": "testbot", "max_parallel_jobs": 3 },
-            "heartbeat": { "enabled": true }
+            "onboard_completed": true
         });
-        std::fs::write(&path, serde_json::to_string_pretty(&legacy).unwrap()).unwrap();
+        std::fs::write(
+            &bootstrap_path,
+            serde_json::to_string_pretty(&bootstrap_json).unwrap(),
+        )
+        .unwrap();
 
-        let config = BootstrapConfig::load_from_legacy(&path);
+        assert!(!env_path.exists());
+        assert!(bootstrap_path.exists());
+
+        // Run the migration
+        migrate_bootstrap_json_to_env(&env_path);
+
+        // .env should now exist with DATABASE_URL
+        assert!(env_path.exists());
+        let content = std::fs::read_to_string(&env_path).unwrap();
         assert_eq!(
-            config.database_url,
-            Some("postgres://localhost/ironclaw".to_string())
+            content,
+            "DATABASE_URL=\"postgres://localhost/ironclaw_upgrade\"\n"
         );
-        assert_eq!(config.database_pool_size, Some(10));
-        assert_eq!(config.secrets_master_key_source, KeySource::Keychain);
-        assert!(config.onboard_completed);
+
+        // bootstrap.json should be renamed to .migrated
+        assert!(!bootstrap_path.exists());
+        assert!(dir.path().join("bootstrap.json.migrated").exists());
     }
 
     #[test]
-    fn test_bootstrap_defaults() {
-        let config = BootstrapConfig::default();
-        assert!(config.database_url.is_none());
-        assert!(config.database_pool_size.is_none());
-        assert_eq!(config.secrets_master_key_source, KeySource::None);
-        assert!(!config.onboard_completed);
+    fn test_migrate_bootstrap_json_no_database_url() {
+        let dir = tempdir().unwrap();
+        let env_path = dir.path().join(".env");
+        let bootstrap_path = dir.path().join("bootstrap.json");
+
+        // bootstrap.json with no database_url
+        let bootstrap_json = serde_json::json!({
+            "onboard_completed": false
+        });
+        std::fs::write(
+            &bootstrap_path,
+            serde_json::to_string_pretty(&bootstrap_json).unwrap(),
+        )
+        .unwrap();
+
+        migrate_bootstrap_json_to_env(&env_path);
+
+        // .env should NOT be created
+        assert!(!env_path.exists());
+        // bootstrap.json should remain (no migration happened)
+        assert!(bootstrap_path.exists());
+    }
+
+    #[test]
+    fn test_migrate_bootstrap_json_missing() {
+        let dir = tempdir().unwrap();
+        let env_path = dir.path().join(".env");
+
+        // No bootstrap.json at all
+        migrate_bootstrap_json_to_env(&env_path);
+
+        // Nothing should happen
+        assert!(!env_path.exists());
+    }
+
+    #[test]
+    fn test_save_bootstrap_env_multiple_vars() {
+        let dir = tempdir().unwrap();
+        let env_path = dir.path().join("nested").join(".env");
+
+        std::fs::create_dir_all(env_path.parent().unwrap()).unwrap();
+
+        let vars = [
+            ("DATABASE_BACKEND", "libsql"),
+            ("LIBSQL_PATH", "/home/user/.ironclaw/ironclaw.db"),
+        ];
+
+        // Write manually to the temp path (save_bootstrap_env uses the global path)
+        let mut content = String::new();
+        for (key, value) in &vars {
+            content.push_str(&format!("{}=\"{}\"\n", key, value));
+        }
+        std::fs::write(&env_path, &content).unwrap();
+
+        // Verify dotenvy can parse all entries
+        let parsed: Vec<(String, String)> = dotenvy::from_path_iter(&env_path)
+            .unwrap()
+            .filter_map(|r| r.ok())
+            .collect();
+        assert_eq!(parsed.len(), 2);
+        assert_eq!(
+            parsed[0],
+            ("DATABASE_BACKEND".to_string(), "libsql".to_string())
+        );
+        assert_eq!(
+            parsed[1],
+            (
+                "LIBSQL_PATH".to_string(),
+                "/home/user/.ironclaw/ironclaw.db".to_string()
+            )
+        );
+    }
+
+    #[test]
+    fn test_save_bootstrap_env_overwrites_previous() {
+        let dir = tempdir().unwrap();
+        let env_path = dir.path().join(".env");
+
+        // Write initial content
+        std::fs::write(&env_path, "DATABASE_URL=\"postgres://old\"\n").unwrap();
+
+        // Overwrite with new vars (simulating save_bootstrap_env behavior)
+        let content = "DATABASE_BACKEND=\"libsql\"\nLIBSQL_PATH=\"/new/path.db\"\n";
+        std::fs::write(&env_path, content).unwrap();
+
+        let parsed: Vec<(String, String)> = dotenvy::from_path_iter(&env_path)
+            .unwrap()
+            .filter_map(|r| r.ok())
+            .collect();
+        // Old DATABASE_URL should be gone
+        assert_eq!(parsed.len(), 2);
+        assert!(parsed.iter().all(|(k, _)| k != "DATABASE_URL"));
     }
 }

src/channels/repl.rs

@@ -33,9 +33,16 @@ use termimad::MadSkin;
 use tokio::sync::mpsc;
 use tokio_stream::wrappers::ReceiverStream;
 
+use crate::agent::truncate_for_preview;
 use crate::channels::{Channel, IncomingMessage, MessageStream, OutgoingResponse, StatusUpdate};
 use crate::error::ChannelError;
 
+/// Max characters for tool result previews in the terminal.
+const CLI_TOOL_RESULT_MAX: usize = 200;
+
+/// Max characters for thinking/status messages in the terminal.
+const CLI_STATUS_MAX: usize = 200;
+
 /// Slash commands available in the REPL.
 const SLASH_COMMANDS: &[&str] = &[
     "/help",
@@ -177,6 +184,8 @@ pub struct ReplChannel {
     debug_mode: Arc<AtomicBool>,
     /// Whether we're currently streaming (chunks have been printed without a trailing newline).
     is_streaming: Arc<AtomicBool>,
+    /// When true, the one-liner startup banner is suppressed (boot screen shown instead).
+    suppress_banner: Arc<AtomicBool>,
 }
 
 impl ReplChannel {
@@ -186,6 +195,7 @@ impl ReplChannel {
             single_message: None,
             debug_mode: Arc::new(AtomicBool::new(false)),
             is_streaming: Arc::new(AtomicBool::new(false)),
+            suppress_banner: Arc::new(AtomicBool::new(false)),
         }
     }
 
@@ -195,9 +205,15 @@ impl ReplChannel {
             single_message: Some(message),
             debug_mode: Arc::new(AtomicBool::new(false)),
             is_streaming: Arc::new(AtomicBool::new(false)),
+            suppress_banner: Arc::new(AtomicBool::new(false)),
         }
     }
 
+    /// Suppress the one-liner startup banner (boot screen will be shown instead).
+    pub fn suppress_banner(&self) {
+        self.suppress_banner.store(true, Ordering::Relaxed);
+    }
+
     fn is_debug(&self) -> bool {
         self.debug_mode.load(Ordering::Relaxed)
     }
@@ -257,11 +273,12 @@ impl Channel for ReplChannel {
         let (tx, rx) = mpsc::channel(32);
         let single_message = self.single_message.clone();
         let debug_mode = Arc::clone(&self.debug_mode);
+        let suppress_banner = Arc::clone(&self.suppress_banner);
 
         std::thread::spawn(move || {
             // Single message mode: send it and return
             if let Some(msg) = single_message {
-                let incoming = IncomingMessage::new("repl", "user", &msg);
+                let incoming = IncomingMessage::new("repl", "default", &msg);
                 let _ = tx.blocking_send(incoming);
                 return;
             }
@@ -291,8 +308,10 @@ impl Channel for ReplChannel {
             }
             let _ = rl.load_history(&hist_path);
 
-            println!("\x1b[1mIronClaw\x1b[0m  /help for commands, /quit to exit");
-            println!();
+            if !suppress_banner.load(Ordering::Relaxed) {
+                println!("\x1b[1mIronClaw\x1b[0m  /help for commands, /quit to exit");
+                println!();
+            }
 
             loop {
                 let prompt = if debug_mode.load(Ordering::Relaxed) {
@@ -329,21 +348,21 @@ impl Channel for ReplChannel {
                             _ => {}
                         }
 
-                        let msg = IncomingMessage::new("repl", "user", line);
+                        let msg = IncomingMessage::new("repl", "default", line);
                         if tx.blocking_send(msg).is_err() {
                             break;
                         }
                     }
                     Err(ReadlineError::Interrupted) => {
                         // Ctrl+C: send /interrupt
-                        let msg = IncomingMessage::new("repl", "user", "/interrupt");
+                        let msg = IncomingMessage::new("repl", "default", "/interrupt");
                         if tx.blocking_send(msg).is_err() {
                             break;
                         }
                     }
                     Err(ReadlineError::Eof) => {
                         // Ctrl+D: send /quit so the agent loop runs graceful shutdown
-                        let msg = IncomingMessage::new("repl", "user", "/quit");
+                        let msg = IncomingMessage::new("repl", "default", "/quit");
                         let _ = tx.blocking_send(msg);
                         break;
                     }
@@ -400,7 +419,8 @@ impl Channel for ReplChannel {
 
         match status {
             StatusUpdate::Thinking(msg) => {
-                eprintln!("  \x1b[90m\u{25CB} {msg}\x1b[0m");
+                let display = truncate_for_preview(&msg, CLI_STATUS_MAX);
+                eprintln!("  \x1b[90m\u{25CB} {display}\x1b[0m");
             }
             StatusUpdate::ToolStarted { name } => {
                 eprintln!("  \x1b[33m\u{25CB} {name}\x1b[0m");
@@ -413,7 +433,8 @@ impl Channel for ReplChannel {
                 }
             }
             StatusUpdate::ToolResult { name: _, preview } => {
-                eprintln!("    \x1b[90m{preview}\x1b[0m");
+                let display = truncate_for_preview(&preview, CLI_TOOL_RESULT_MAX);
+                eprintln!("    \x1b[90m{display}\x1b[0m");
             }
             StatusUpdate::StreamChunk(chunk) => {
                 // Print separator on the false-to-true transition
@@ -438,7 +459,8 @@ impl Channel for ReplChannel {
             }
             StatusUpdate::Status(msg) => {
                 if debug || msg.contains("approval") || msg.contains("Approval") {
-                    eprintln!("  \x1b[90m{msg}\x1b[0m");
+                    let display = truncate_for_preview(&msg, CLI_STATUS_MAX);
+                    eprintln!("  \x1b[90m{display}\x1b[0m");
                 }
             }
             StatusUpdate::ApprovalNeeded {

src/channels/wasm/wrapper.rs

@@ -76,6 +76,9 @@ struct ChannelStoreData {
     credentials: HashMap<String, String>,
     /// Pairing store for DM pairing (guest access control).
     pairing_store: Arc<PairingStore>,
+    /// Dedicated tokio runtime for HTTP requests, lazily initialized.
+    /// Reused across multiple `http_request` calls within one execution.
+    http_runtime: Option<tokio::runtime::Runtime>,
 }
 
 impl ChannelStoreData {
@@ -96,6 +99,7 @@ impl ChannelStoreData {
             table: ResourceTable::new(),
             credentials,
             pairing_store,
+            http_runtime: None,
         }
     }
 
@@ -134,13 +138,13 @@ impl ChannelStoreData {
         if result.contains('{') && result.contains('}') {
             // Only warn if it looks like an unresolved placeholder (not JSON braces)
             let brace_pattern = regex::Regex::new(r"\{[A-Z_]+\}").ok();
-            if let Some(re) = brace_pattern {
-                if re.is_match(&result) {
-                    tracing::warn!(
-                        context = %context,
-                        "String may contain unresolved credential placeholders"
-                    );
-                }
+            if let Some(re) = brace_pattern
+                && re.is_match(&result)
+            {
+                tracing::warn!(
+                    context = %context,
+                    "String may contain unresolved credential placeholders"
+                );
             }
         }
 
@@ -273,10 +277,35 @@ impl near::agent::channel_host::Host for ChannelStoreData {
             .scan_http_request(&url, &header_vec, body.as_deref())
             .map_err(|e| format!("Potential secret leak blocked: {}", e))?;
 
-        // Make the HTTP request using blocking I/O
-        // We're already in a spawn_blocking context, so we can use block_on
-        let result = tokio::runtime::Handle::current().block_on(async {
-            let client = reqwest::Client::new();
+        // Get the max response size from capabilities (default 10MB).
+        let max_response_bytes = self
+            .host_state
+            .capabilities()
+            .tool_capabilities
+            .http
+            .as_ref()
+            .map(|h| h.max_response_bytes)
+            .unwrap_or(10 * 1024 * 1024);
+
+        // Make the HTTP request using a dedicated single-threaded runtime.
+        // We're inside spawn_blocking, so we can't rely on the main runtime's
+        // I/O driver (it may be busy with WASM compilation or other startup work).
+        // A dedicated runtime gives us our own I/O driver and avoids contention.
+        // The runtime is lazily created and reused across calls within one execution.
+        if self.http_runtime.is_none() {
+            self.http_runtime = Some(
+                tokio::runtime::Builder::new_current_thread()
+                    .enable_all()
+                    .build()
+                    .map_err(|e| format!("Failed to create HTTP runtime: {e}"))?,
+            );
+        }
+        let rt = self.http_runtime.as_ref().expect("just initialized");
+        let result = rt.block_on(async {
+            let client = reqwest::Client::builder()
+                .connect_timeout(std::time::Duration::from_secs(10))
+                .build()
+                .map_err(|e| format!("Failed to build HTTP client: {e}"))?;
 
             let mut request = match method.to_uppercase().as_str() {
                 "GET" => client.get(&url),
@@ -298,9 +327,9 @@ impl near::agent::channel_host::Host for ChannelStoreData {
                 request = request.body(body_bytes);
             }
 
-            // Send request with caller-specified timeout (default 30s).
-            // Cap at callback_timeout to prevent outliving the host wrapper.
-            let timeout = std::time::Duration::from_millis(timeout_ms.unwrap_or(30_000) as u64);
+            // Send request with caller-specified timeout (default 30s, max 5min).
+            let timeout_ms = timeout_ms.unwrap_or(30_000).min(300_000) as u64;
+            let timeout = std::time::Duration::from_millis(timeout_ms);
             let response = request.timeout(timeout).send().await.map_err(|e| {
                 // Walk the full error chain so we get the actual root cause
                 // (DNS, TLS, connection refused, etc.) instead of just
@@ -325,11 +354,29 @@ impl near::agent::channel_host::Host for ChannelStoreData {
                 })
                 .collect();
             let headers_json = serde_json::to_string(&response_headers).unwrap_or_default();
+
+            // Enforce max response body size to prevent memory exhaustion.
+            let max_response = max_response_bytes;
+            if let Some(cl) = response.content_length()
+                && cl as usize > max_response
+            {
+                return Err(format!(
+                    "Response body too large: {} bytes exceeds limit of {} bytes",
+                    cl, max_response
+                ));
+            }
             let body = response
                 .bytes()
                 .await
-                .map_err(|e| format!("Failed to read response body: {}", e))?
-                .to_vec();
+                .map_err(|e| format!("Failed to read response body: {}", e))?;
+            if body.len() > max_response {
+                return Err(format!(
+                    "Response body too large: {} bytes exceeds limit of {} bytes",
+                    body.len(),
+                    max_response
+                ));
+            }
+            let body = body.to_vec();
 
             tracing::info!(
                 status = status,
@@ -767,7 +814,21 @@ impl WasmChannel {
         .await;
 
         match result {
-            Ok(Ok((config, _host_state))) => {
+            Ok(Ok((config, mut host_state))) => {
+                // Surface WASM guest logs (errors/warnings from webhook setup, etc.)
+                for entry in host_state.take_logs() {
+                    match entry.level {
+                        crate::tools::wasm::LogLevel::Error => {
+                            tracing::error!(channel = %self.name, "{}", entry.message);
+                        }
+                        crate::tools::wasm::LogLevel::Warn => {
+                            tracing::warn!(channel = %self.name, "{}", entry.message);
+                        }
+                        _ => {
+                            tracing::debug!(channel = %self.name, "{}", entry.message);
+                        }
+                    }
+                }
                 tracing::info!(
                     channel = %self.name,
                     display_name = %config.display_name,
@@ -1467,8 +1528,8 @@ impl WasmChannel {
                         match result {
                             Ok(emitted_messages) => {
                                 // Process any emitted messages
-                                if !emitted_messages.is_empty() {
-                                    if let Err(e) = Self::dispatch_emitted_messages(
+                                if !emitted_messages.is_empty()
+                                    && let Err(e) = Self::dispatch_emitted_messages(
                                         &channel_name,
                                         emitted_messages,
                                         &message_tx,
@@ -1480,7 +1541,6 @@ impl WasmChannel {
                                             "Failed to dispatch emitted messages from poll"
                                         );
                                     }
-                                }
                             }
                             Err(e) => {
                                 tracing::warn!(
@@ -1710,22 +1770,22 @@ impl Channel for WasmChannel {
         *self.endpoints.write().await = endpoints;
 
         // Start polling if configured
-        if let Some(poll_config) = &config.poll {
-            if poll_config.enabled {
-                let interval = self
-                    .capabilities
-                    .validate_poll_interval(poll_config.interval_ms)
-                    .map_err(|e| ChannelError::StartupFailed {
-                        name: self.name.clone(),
-                        reason: e,
-                    })?;
+        if let Some(poll_config) = &config.poll
+            && poll_config.enabled
+        {
+            let interval = self
+                .capabilities
+                .validate_poll_interval(poll_config.interval_ms)
+                .map_err(|e| ChannelError::StartupFailed {
+                    name: self.name.clone(),
+                    reason: e,
+                })?;
 
-                // Create shutdown channel for polling and store the sender to keep it alive
-                let (poll_shutdown_tx, poll_shutdown_rx) = oneshot::channel();
-                *self.poll_shutdown_tx.write().await = Some(poll_shutdown_tx);
+            // Create shutdown channel for polling and store the sender to keep it alive
+            let (poll_shutdown_tx, poll_shutdown_rx) = oneshot::channel();
+            *self.poll_shutdown_tx.write().await = Some(poll_shutdown_tx);
 
-                self.start_polling(Duration::from_millis(interval as u64), poll_shutdown_rx);
-            }
+            self.start_polling(Duration::from_millis(interval as u64), poll_shutdown_rx);
         }
 
         tracing::info!(
@@ -2588,15 +2648,52 @@ mod tests {
         assert_eq!(store.redact_credentials(input), input);
     }
 
-    /// Verify that the block_on-inside-spawn_blocking pattern used by the WASM
-    /// channel HTTP host function doesn't deadlock or panic.
+    /// Verify that WASM HTTP host functions work using a dedicated
+    /// current-thread runtime inside spawn_blocking.
     #[tokio::test]
-    async fn test_block_on_inside_spawn_blocking_does_not_deadlock() {
+    async fn test_dedicated_runtime_inside_spawn_blocking() {
         let result = tokio::task::spawn_blocking(|| {
-            tokio::runtime::Handle::current().block_on(async { 42 })
+            let rt = tokio::runtime::Builder::new_current_thread()
+                .enable_all()
+                .build()
+                .expect("failed to build runtime");
+            rt.block_on(async { 42 })
         })
         .await
         .expect("spawn_blocking panicked");
         assert_eq!(result, 42);
     }
+
+    /// Verify a real HTTP request works using the dedicated-runtime pattern.
+    /// This catches DNS, TLS, and I/O driver issues that trivial tests miss.
+    #[tokio::test]
+    #[ignore] // requires network
+    async fn test_dedicated_runtime_real_http() {
+        let result = tokio::task::spawn_blocking(|| {
+            let rt = tokio::runtime::Builder::new_current_thread()
+                .enable_all()
+                .build()
+                .expect("failed to build runtime");
+            rt.block_on(async {
+                let client = reqwest::Client::builder()
+                    .connect_timeout(std::time::Duration::from_secs(10))
+                    .build()
+                    .expect("failed to build client");
+                let resp = client
+                    .get("https://api.telegram.org/bot000/getMe")
+                    .timeout(std::time::Duration::from_secs(10))
+                    .send()
+                    .await;
+                match resp {
+                    Ok(r) => r.status().as_u16(),
+                    Err(e) if e.is_timeout() => panic!("request timed out: {e}"),
+                    Err(e) => panic!("unexpected error: {e}"),
+                }
+            })
+        })
+        .await
+        .expect("spawn_blocking panicked");
+        // 404 because "000" is not a valid bot token
+        assert_eq!(result, 404);
+    }
 }

src/channels/web/auth.rs

@@ -6,6 +6,7 @@ use axum::{
     middleware::Next,
     response::{IntoResponse, Response},
 };
+use subtle::ConstantTimeEq;
 
 /// Shared auth state injected via axum middleware state.
 #[derive(Clone)]
@@ -23,24 +24,22 @@ pub async fn auth_middleware(
     request: Request,
     next: Next,
 ) -> Response {
-    // Try Authorization header first
-    if let Some(auth_header) = headers.get("authorization") {
-        if let Ok(value) = auth_header.to_str() {
-            if let Some(token) = value.strip_prefix("Bearer ") {
-                if token == auth.token {
-                    return next.run(request).await;
-                }
-            }
-        }
+    // Try Authorization header first (constant-time comparison)
+    if let Some(auth_header) = headers.get("authorization")
+        && let Ok(value) = auth_header.to_str()
+        && let Some(token) = value.strip_prefix("Bearer ")
+        && bool::from(token.as_bytes().ct_eq(auth.token.as_bytes()))
+    {
+        return next.run(request).await;
     }
 
-    // Fall back to query parameter (for SSE EventSource)
+    // Fall back to query parameter for SSE EventSource (constant-time comparison)
     if let Some(query) = request.uri().query() {
         for pair in query.split('&') {
-            if let Some(token) = pair.strip_prefix("token=") {
-                if token == auth.token {
-                    return next.run(request).await;
-                }
+            if let Some(token) = pair.strip_prefix("token=")
+                && bool::from(token.as_bytes().ct_eq(auth.token.as_bytes()))
+            {
+                return next.run(request).await;
             }
         }
     }

src/channels/web/log_layer.rs

@@ -24,6 +24,8 @@ use tokio::sync::broadcast;
 use tracing::field::{Field, Visit};
 use tracing_subscriber::Layer;
 
+use crate::safety::LeakDetector;
+
 /// Maximum number of recent log entries kept for late-joining SSE subscribers.
 const HISTORY_CAP: usize = 500;
 
@@ -46,6 +48,8 @@ pub struct LogEntry {
 pub struct LogBroadcaster {
     tx: broadcast::Sender<LogEntry>,
     recent: Mutex<VecDeque<LogEntry>>,
+    /// Scrubs secrets from log messages before broadcasting to SSE clients.
+    leak_detector: LeakDetector,
 }
 
 impl LogBroadcaster {
@@ -54,10 +58,19 @@ impl LogBroadcaster {
         Self {
             tx,
             recent: Mutex::new(VecDeque::with_capacity(HISTORY_CAP)),
+            leak_detector: LeakDetector::new(),
         }
     }
 
-    pub fn send(&self, entry: LogEntry) {
+    pub fn send(&self, mut entry: LogEntry) {
+        // Scrub secrets from the message before it reaches any subscriber.
+        // This is defense-in-depth: even if code elsewhere accidentally logs
+        // a secret, it won't be broadcast to SSE clients.
+        entry.message = self
+            .leak_detector
+            .scan_and_clean(&entry.message)
+            .unwrap_or_else(|_| "[log message redacted: contained blocked secret]".to_string());
+
         // Stash in ring buffer (for late joiners)
         if let Ok(mut buf) = self.recent.lock() {
             if buf.len() >= HISTORY_CAP {
@@ -145,6 +158,9 @@ impl Visit for MessageVisitor {
 ///
 /// Only forwards DEBUG and above. Attach to the tracing subscriber
 /// alongside the existing fmt layer.
+///
+/// Log messages are scrubbed through `LeakDetector` in `LogBroadcaster::send()`
+/// (the single funnel point for all log output, including late-joiner history).
 pub struct WebLogLayer {
     broadcaster: Arc<LogBroadcaster>,
 }
@@ -178,6 +194,7 @@ impl<S: tracing::Subscriber> Layer<S> for WebLogLayer {
             timestamp: chrono::Utc::now().to_rfc3339_opts(chrono::SecondsFormat::Millis, true),
         };
 
+        // LeakDetector scrubbing happens inside broadcaster.send()
         self.broadcaster.send(entry);
     }
 }
@@ -313,4 +330,29 @@ mod tests {
         let v = MessageVisitor::new();
         assert_eq!(v.finish(), "");
     }
+
+    #[test]
+    fn test_broadcaster_has_leak_detector() {
+        let broadcaster = LogBroadcaster::new();
+        // Verify the leak detector is initialized with default patterns
+        assert!(broadcaster.leak_detector.pattern_count() > 0);
+    }
+
+    #[test]
+    fn test_leak_detector_scrubs_api_key_in_log() {
+        let detector = crate::safety::LeakDetector::new();
+        let msg = "Connecting with token sk-proj-test1234567890abcdefghij";
+        let result = detector.scan_and_clean(msg);
+        // Should be blocked (OpenAI key pattern)
+        assert!(result.is_err());
+    }
+
+    #[test]
+    fn test_leak_detector_passes_clean_log() {
+        let detector = crate::safety::LeakDetector::new();
+        let msg = "Request completed status=200 url=https://api.example.com/data";
+        let result = detector.scan_and_clean(msg);
+        assert!(result.is_ok());
+        assert_eq!(result.unwrap(), msg);
+    }
 }

src/channels/web/mod.rs

@@ -16,6 +16,7 @@
 
 pub mod auth;
 pub mod log_layer;
+pub mod openai_compat;
 pub mod server;
 pub mod sse;
 pub mod types;
@@ -31,9 +32,9 @@ use tokio_stream::wrappers::ReceiverStream;
 use crate::agent::SessionManager;
 use crate::channels::{Channel, IncomingMessage, MessageStream, OutgoingResponse, StatusUpdate};
 use crate::config::GatewayConfig;
+use crate::db::Database;
 use crate::error::ChannelError;
 use crate::extensions::ExtensionManager;
-use crate::history::Store;
 use crate::orchestrator::job_manager::ContainerJobManager;
 use crate::tools::ToolRegistry;
 use crate::workspace::Workspace;
@@ -81,6 +82,8 @@ impl GatewayChannel {
             user_id: config.user_id.clone(),
             shutdown_tx: tokio::sync::RwLock::new(None),
             ws_tracker: Some(Arc::new(ws::WsConnectionTracker::new())),
+            llm_provider: None,
+            chat_rate_limiter: server::RateLimiter::new(30, 60),
         });
 
         Self {
@@ -106,6 +109,8 @@ impl GatewayChannel {
             user_id: self.state.user_id.clone(),
             shutdown_tx: tokio::sync::RwLock::new(None),
             ws_tracker: self.state.ws_tracker.clone(),
+            llm_provider: self.state.llm_provider.clone(),
+            chat_rate_limiter: server::RateLimiter::new(30, 60),
         };
         mutate(&mut new_state);
         self.state = Arc::new(new_state);
@@ -142,7 +147,7 @@ impl GatewayChannel {
     }
 
     /// Inject the database store for sandbox job persistence.
-    pub fn with_store(mut self, store: Arc<Store>) -> Self {
+    pub fn with_store(mut self, store: Arc<dyn Database>) -> Self {
         self.rebuild_state(|s| s.store = Some(store));
         self
     }
@@ -169,6 +174,12 @@ impl GatewayChannel {
         self
     }
 
+    /// Inject the LLM provider for OpenAI-compatible API proxy.
+    pub fn with_llm_provider(mut self, llm: Arc<dyn crate::llm::LlmProvider>) -> Self {
+        self.rebuild_state(|s| s.llm_provider = Some(llm));
+        self
+    }
+
     /// Get the auth token (for printing to console on startup).
     pub fn auth_token(&self) -> &str {
         &self.auth_token

src/channels/web/server.rs

@@ -5,10 +5,11 @@
 use std::convert::Infallible;
 use std::net::SocketAddr;
 use std::sync::Arc;
+use std::sync::atomic::{AtomicU64, Ordering};
 
 use axum::{
     Json, Router,
-    extract::{Path, Query, State, WebSocketUpgrade},
+    extract::{DefaultBodyLimit, Path, Query, State, WebSocketUpgrade},
     http::{StatusCode, header},
     middleware,
     response::{
@@ -20,6 +21,7 @@ use axum::{
 use serde::Deserialize;
 use tokio::sync::{mpsc, oneshot};
 use tokio_stream::StreamExt;
+use tower_http::cors::{AllowHeaders, CorsLayer};
 use uuid::Uuid;
 
 use crate::agent::SessionManager;
@@ -28,8 +30,8 @@ use crate::channels::web::auth::{AuthState, auth_middleware};
 use crate::channels::web::log_layer::LogBroadcaster;
 use crate::channels::web::sse::SseManager;
 use crate::channels::web::types::*;
+use crate::db::Database;
 use crate::extensions::ExtensionManager;
-use crate::history::Store;
 use crate::orchestrator::job_manager::ContainerJobManager;
 use crate::tools::ToolRegistry;
 use crate::workspace::Workspace;
@@ -44,6 +46,69 @@ pub type PromptQueue = Arc<
     >,
 >;
 
+/// Simple sliding-window rate limiter.
+///
+/// Tracks the number of requests in the current window. Resets when the window expires.
+/// Not per-IP (since this is a single-user gateway with auth), but prevents flooding.
+pub struct RateLimiter {
+    /// Requests remaining in the current window.
+    remaining: AtomicU64,
+    /// Epoch second when the current window started.
+    window_start: AtomicU64,
+    /// Maximum requests per window.
+    max_requests: u64,
+    /// Window duration in seconds.
+    window_secs: u64,
+}
+
+impl RateLimiter {
+    pub fn new(max_requests: u64, window_secs: u64) -> Self {
+        Self {
+            remaining: AtomicU64::new(max_requests),
+            window_start: AtomicU64::new(
+                std::time::SystemTime::now()
+                    .duration_since(std::time::UNIX_EPOCH)
+                    .unwrap_or_default()
+                    .as_secs(),
+            ),
+            max_requests,
+            window_secs,
+        }
+    }
+
+    /// Try to consume one request. Returns `true` if allowed, `false` if rate limited.
+    pub fn check(&self) -> bool {
+        let now = std::time::SystemTime::now()
+            .duration_since(std::time::UNIX_EPOCH)
+            .unwrap_or_default()
+            .as_secs();
+
+        let window = self.window_start.load(Ordering::Relaxed);
+        if now.saturating_sub(window) >= self.window_secs {
+            // Window expired, reset
+            self.window_start.store(now, Ordering::Relaxed);
+            self.remaining
+                .store(self.max_requests - 1, Ordering::Relaxed);
+            return true;
+        }
+
+        // Try to decrement remaining
+        loop {
+            let current = self.remaining.load(Ordering::Relaxed);
+            if current == 0 {
+                return false;
+            }
+            if self
+                .remaining
+                .compare_exchange_weak(current, current - 1, Ordering::Relaxed, Ordering::Relaxed)
+                .is_ok()
+            {
+                return true;
+            }
+        }
+    }
+}
+
 /// Shared state for all gateway handlers.
 pub struct GatewayState {
     /// Channel to send messages to the agent loop.
@@ -61,7 +126,7 @@ pub struct GatewayState {
     /// Tool registry for listing registered tools.
     pub tool_registry: Option<Arc<ToolRegistry>>,
     /// Database store for sandbox job persistence.
-    pub store: Option<Arc<Store>>,
+    pub store: Option<Arc<dyn Database>>,
     /// Container job manager for sandbox operations.
     pub job_manager: Option<Arc<ContainerJobManager>>,
     /// Prompt queue for Claude Code follow-up prompts.
@@ -72,6 +137,10 @@ pub struct GatewayState {
     pub shutdown_tx: tokio::sync::RwLock<Option<oneshot::Sender<()>>>,
     /// WebSocket connection tracker.
     pub ws_tracker: Option<Arc<crate::channels::web::ws::WsConnectionTracker>>,
+    /// LLM provider for OpenAI-compatible API proxy.
+    pub llm_provider: Option<Arc<dyn crate::llm::LlmProvider>>,
+    /// Rate limiter for chat endpoints (30 messages per 60 seconds).
+    pub chat_rate_limiter: RateLimiter,
 }
 
 /// Start the gateway HTTP server.
@@ -168,7 +237,16 @@ pub async fn start_server(
         )
         // Gateway control plane
         .route("/api/gateway/status", get(gateway_status_handler))
-        .route_layer(middleware::from_fn_with_state(auth_state, auth_middleware));
+        // OpenAI-compatible API
+        .route(
+            "/v1/chat/completions",
+            post(super::openai_compat::chat_completions_handler),
+        )
+        .route("/v1/models", get(super::openai_compat::models_handler))
+        .route_layer(middleware::from_fn_with_state(
+            auth_state.clone(),
+            auth_middleware,
+        ));
 
     // Static file routes (no auth, served from embedded strings)
     let statics = Router::new()
@@ -176,19 +254,46 @@ pub async fn start_server(
         .route("/style.css", get(css_handler))
         .route("/app.js", get(js_handler));
 
-    // Project file serving (no auth, local browsing of sandbox outputs).
-    // The trailing-slash route serves index.html; the bare route redirects so
-    // relative paths in the HTML (e.g. href="style.css") resolve correctly.
+    // Project file serving (behind auth to prevent unauthorized file access).
     let projects = Router::new()
         .route("/projects/{project_id}", get(project_redirect_handler))
         .route("/projects/{project_id}/", get(project_index_handler))
-        .route("/projects/{project_id}/{*path}", get(project_file_handler));
+        .route("/projects/{project_id}/{*path}", get(project_file_handler))
+        .route_layer(middleware::from_fn_with_state(
+            auth_state.clone(),
+            auth_middleware,
+        ));
+
+    // CORS: restrict to same-origin by default. Only localhost/127.0.0.1
+    // origins are allowed, since the gateway is a local-first service.
+    let cors = CorsLayer::new()
+        .allow_origin([
+            format!("http://{}:{}", addr.ip(), addr.port())
+                .parse()
+                .expect("valid origin"),
+            format!("http://localhost:{}", addr.port())
+                .parse()
+                .expect("valid origin"),
+        ])
+        .allow_methods([
+            axum::http::Method::GET,
+            axum::http::Method::POST,
+            axum::http::Method::PUT,
+            axum::http::Method::DELETE,
+        ])
+        .allow_headers(AllowHeaders::list([
+            header::CONTENT_TYPE,
+            header::AUTHORIZATION,
+        ]))
+        .allow_credentials(true);
 
     let app = Router::new()
         .merge(public)
         .merge(statics)
         .merge(projects)
         .merge(protected)
+        .layer(cors)
+        .layer(DefaultBodyLimit::max(1024 * 1024)) // 1 MB max request body
         .with_state(state.clone());
 
     let (shutdown_tx, shutdown_rx) = oneshot::channel();
@@ -244,6 +349,13 @@ async fn chat_send_handler(
     State(state): State<Arc<GatewayState>>,
     Json(req): Json<SendMessageRequest>,
 ) -> Result<(StatusCode, Json<SendMessageResponse>), (StatusCode, String)> {
+    if !state.chat_rate_limiter.check() {
+        return Err((
+            StatusCode::TOO_MANY_REQUESTS,
+            "Rate limit exceeded. Try again shortly.".to_string(),
+        ));
+    }
+
     let mut msg = IncomingMessage::new("gateway", &state.user_id, &req.content);
 
     if let Some(ref thread_id) = req.thread_id {
@@ -413,24 +525,58 @@ pub async fn clear_auth_mode(state: &GatewayState) {
     if let Some(ref sm) = state.session_manager {
         let session = sm.get_or_create_session(&state.user_id).await;
         let mut sess = session.lock().await;
-        if let Some(thread_id) = sess.active_thread {
-            if let Some(thread) = sess.threads.get_mut(&thread_id) {
-                thread.pending_auth = None;
-            }
+        if let Some(thread_id) = sess.active_thread
+            && let Some(thread) = sess.threads.get_mut(&thread_id)
+        {
+            thread.pending_auth = None;
         }
     }
 }
 
-async fn chat_events_handler(State(state): State<Arc<GatewayState>>) -> impl IntoResponse {
-    // subscribe() returns Sse<impl Stream + 'static + use<>> so no lifetime issues
-    state.sse.subscribe()
+async fn chat_events_handler(
+    State(state): State<Arc<GatewayState>>,
+) -> Result<impl IntoResponse, (StatusCode, String)> {
+    state.sse.subscribe().ok_or((
+        StatusCode::SERVICE_UNAVAILABLE,
+        "Too many connections".to_string(),
+    ))
 }
 
 async fn chat_ws_handler(
+    headers: axum::http::HeaderMap,
     ws: WebSocketUpgrade,
     State(state): State<Arc<GatewayState>>,
-) -> impl IntoResponse {
-    ws.on_upgrade(move |socket| crate::channels::web::ws::handle_ws_connection(socket, state))
+) -> Result<impl IntoResponse, (StatusCode, String)> {
+    // Validate Origin header to prevent cross-site WebSocket hijacking.
+    // Require the header outright; browsers always send it for WS upgrades,
+    // so a missing Origin means a non-browser client trying to bypass the check.
+    let origin = headers
+        .get("origin")
+        .and_then(|v| v.to_str().ok())
+        .ok_or_else(|| {
+            (
+                StatusCode::FORBIDDEN,
+                "WebSocket Origin header required".to_string(),
+            )
+        })?;
+
+    // Extract the host from the origin and compare exactly, so that
+    // crafted origins like "http://localhost.evil.com" are rejected.
+    // Origin format is "scheme://host[:port]".
+    let host = origin
+        .strip_prefix("http://")
+        .or_else(|| origin.strip_prefix("https://"))
+        .and_then(|rest| rest.split(':').next()?.split('/').next())
+        .unwrap_or("");
+
+    let is_local = matches!(host, "localhost" | "127.0.0.1" | "[::1]");
+    if !is_local {
+        return Err((
+            StatusCode::FORBIDDEN,
+            "WebSocket origin not allowed".to_string(),
+        ));
+    }
+    Ok(ws.on_upgrade(move |socket| crate::channels::web::ws::handle_ws_connection(socket, state)))
 }
 
 #[derive(Deserialize)]
@@ -477,57 +623,72 @@ async fn chat_history_handler(
             .ok_or((StatusCode::NOT_FOUND, "No active thread".to_string()))?
     };
 
-    // For paginated requests (before cursor set), always go to DB
-    if before_cursor.is_some() {
-        if let Some(ref store) = state.store {
-            let (messages, has_more) = store
-                .list_conversation_messages_paginated(thread_id, before_cursor, limit as i64)
-                .await
-                .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
-
-            let oldest_timestamp = messages.first().map(|m| m.created_at.to_rfc3339());
-            let turns = build_turns_from_db_messages(&messages);
-            return Ok(Json(HistoryResponse {
-                thread_id,
-                turns,
-                has_more,
-                oldest_timestamp,
-            }));
+    // Verify the thread belongs to the authenticated user before returning any data.
+    // In-memory threads are already scoped by user via session_manager, but DB
+    // lookups could expose another user's conversation if the UUID is guessed.
+    if query.thread_id.is_some()
+        && let Some(ref store) = state.store
+    {
+        let owned = store
+            .conversation_belongs_to_user(thread_id, &state.user_id)
+            .await
+            .unwrap_or(false);
+        if !owned && !sess.threads.contains_key(&thread_id) {
+            return Err((StatusCode::NOT_FOUND, "Thread not found".to_string()));
         }
     }
 
-    // Try in-memory first (freshest data for active threads)
-    if let Some(thread) = sess.threads.get(&thread_id) {
-        if !thread.turns.is_empty() {
-            let turns: Vec<TurnInfo> = thread
-                .turns
-                .iter()
-                .map(|t| TurnInfo {
-                    turn_number: t.turn_number,
-                    user_input: t.user_input.clone(),
-                    response: t.response.clone(),
-                    state: format!("{:?}", t.state),
-                    started_at: t.started_at.to_rfc3339(),
-                    completed_at: t.completed_at.map(|dt| dt.to_rfc3339()),
-                    tool_calls: t
-                        .tool_calls
-                        .iter()
-                        .map(|tc| ToolCallInfo {
-                            name: tc.name.clone(),
-                            has_result: tc.result.is_some(),
-                            has_error: tc.error.is_some(),
-                        })
-                        .collect(),
-                })
-                .collect();
+    // For paginated requests (before cursor set), always go to DB
+    if before_cursor.is_some()
+        && let Some(ref store) = state.store
+    {
+        let (messages, has_more) = store
+            .list_conversation_messages_paginated(thread_id, before_cursor, limit as i64)
+            .await
+            .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
 
-            return Ok(Json(HistoryResponse {
-                thread_id,
-                turns,
-                has_more: false,
-                oldest_timestamp: None,
-            }));
-        }
+        let oldest_timestamp = messages.first().map(|m| m.created_at.to_rfc3339());
+        let turns = build_turns_from_db_messages(&messages);
+        return Ok(Json(HistoryResponse {
+            thread_id,
+            turns,
+            has_more,
+            oldest_timestamp,
+        }));
+    }
+
+    // Try in-memory first (freshest data for active threads)
+    if let Some(thread) = sess.threads.get(&thread_id)
+        && !thread.turns.is_empty()
+    {
+        let turns: Vec<TurnInfo> = thread
+            .turns
+            .iter()
+            .map(|t| TurnInfo {
+                turn_number: t.turn_number,
+                user_input: t.user_input.clone(),
+                response: t.response.clone(),
+                state: format!("{:?}", t.state),
+                started_at: t.started_at.to_rfc3339(),
+                completed_at: t.completed_at.map(|dt| dt.to_rfc3339()),
+                tool_calls: t
+                    .tool_calls
+                    .iter()
+                    .map(|tc| ToolCallInfo {
+                        name: tc.name.clone(),
+                        has_result: tc.result.is_some(),
+                        has_error: tc.error.is_some(),
+                    })
+                    .collect(),
+            })
+            .collect();
+
+        return Ok(Json(HistoryResponse {
+            thread_id,
+            turns,
+            has_more: false,
+            oldest_timestamp: None,
+        }));
     }
 
     // Fall back to DB for historical threads not in memory (paginated)
@@ -577,12 +738,12 @@ fn build_turns_from_db_messages(messages: &[crate::history::ConversationMessage]
             };
 
             // Check if next message is an assistant response
-            if let Some(next) = iter.peek() {
-                if next.role == "assistant" {
-                    let assistant_msg = iter.next().expect("peeked");
-                    turn.response = Some(assistant_msg.content.clone());
-                    turn.completed_at = Some(assistant_msg.created_at.to_rfc3339());
-                }
+            if let Some(next) = iter.peek()
+                && next.role == "assistant"
+            {
+                let assistant_msg = iter.next().expect("peeked");
+                turn.response = Some(assistant_msg.content.clone());
+                turn.completed_at = Some(assistant_msg.created_at.to_rfc3339());
             }
 
             // Incomplete turn (user message without response)
@@ -901,14 +1062,16 @@ async fn jobs_list_handler(
         "Database not available".to_string(),
     ))?;
 
-    // Fetch sandbox jobs from the DB.
+    // Fetch sandbox jobs scoped to the authenticated user.
     let sandbox_jobs = store
-        .list_sandbox_jobs()
+        .list_sandbox_jobs_for_user(&state.user_id)
         .await
         .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
 
+    // Scope jobs to the authenticated user.
     let mut jobs: Vec<JobInfo> = sandbox_jobs
         .iter()
+        .filter(|j| j.user_id == state.user_id)
         .map(|j| {
             let ui_state = match j.status.as_str() {
                 "creating" => "pending",
@@ -941,7 +1104,7 @@ async fn jobs_summary_handler(
     ))?;
 
     let s = store
-        .sandbox_job_summary()
+        .sandbox_job_summary_for_user(&state.user_id)
         .await
         .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
 
@@ -962,63 +1125,66 @@ async fn jobs_detail_handler(
     let job_id = Uuid::parse_str(&id)
         .map_err(|_| (StatusCode::BAD_REQUEST, "Invalid job ID".to_string()))?;
 
-    // Try sandbox job from DB first.
-    if let Some(ref store) = state.store {
-        if let Ok(Some(job)) = store.get_sandbox_job(job_id).await {
-            let browse_id = std::path::Path::new(&job.project_dir)
-                .file_name()
-                .map(|n| n.to_string_lossy().to_string())
-                .unwrap_or_else(|| job.id.to_string());
-
-            let ui_state = match job.status.as_str() {
-                "creating" => "pending",
-                "running" => "in_progress",
-                s => s,
-            };
-
-            let elapsed_secs = job.started_at.map(|start| {
-                let end = job.completed_at.unwrap_or_else(chrono::Utc::now);
-                (end - start).num_seconds().max(0) as u64
-            });
-
-            // Synthesize transitions from timestamps.
-            let mut transitions = Vec::new();
-            if let Some(started) = job.started_at {
-                transitions.push(TransitionInfo {
-                    from: "creating".to_string(),
-                    to: "running".to_string(),
-                    timestamp: started.to_rfc3339(),
-                    reason: None,
-                });
-            }
-            if let Some(completed) = job.completed_at {
-                transitions.push(TransitionInfo {
-                    from: "running".to_string(),
-                    to: job.status.clone(),
-                    timestamp: completed.to_rfc3339(),
-                    reason: job.failure_reason.clone(),
-                });
-            }
-
-            return Ok(Json(JobDetailResponse {
-                id: job.id,
-                title: job.task.clone(),
-                description: String::new(),
-                state: ui_state.to_string(),
-                user_id: job.user_id.clone(),
-                created_at: job.created_at.to_rfc3339(),
-                started_at: job.started_at.map(|dt| dt.to_rfc3339()),
-                completed_at: job.completed_at.map(|dt| dt.to_rfc3339()),
-                elapsed_secs,
-                project_dir: Some(job.project_dir.clone()),
-                browse_url: Some(format!("/projects/{}/", browse_id)),
-                job_mode: {
-                    let mode = store.get_sandbox_job_mode(job.id).await.ok().flatten();
-                    mode.filter(|m| m != "worker")
-                },
-                transitions,
-            }));
+    // Try sandbox job from DB first, scoped to the authenticated user.
+    if let Some(ref store) = state.store
+        && let Ok(Some(job)) = store.get_sandbox_job(job_id).await
+    {
+        if job.user_id != state.user_id {
+            return Err((StatusCode::NOT_FOUND, "Job not found".to_string()));
         }
+        let browse_id = std::path::Path::new(&job.project_dir)
+            .file_name()
+            .map(|n| n.to_string_lossy().to_string())
+            .unwrap_or_else(|| job.id.to_string());
+
+        let ui_state = match job.status.as_str() {
+            "creating" => "pending",
+            "running" => "in_progress",
+            s => s,
+        };
+
+        let elapsed_secs = job.started_at.map(|start| {
+            let end = job.completed_at.unwrap_or_else(chrono::Utc::now);
+            (end - start).num_seconds().max(0) as u64
+        });
+
+        // Synthesize transitions from timestamps.
+        let mut transitions = Vec::new();
+        if let Some(started) = job.started_at {
+            transitions.push(TransitionInfo {
+                from: "creating".to_string(),
+                to: "running".to_string(),
+                timestamp: started.to_rfc3339(),
+                reason: None,
+            });
+        }
+        if let Some(completed) = job.completed_at {
+            transitions.push(TransitionInfo {
+                from: "running".to_string(),
+                to: job.status.clone(),
+                timestamp: completed.to_rfc3339(),
+                reason: job.failure_reason.clone(),
+            });
+        }
+
+        return Ok(Json(JobDetailResponse {
+            id: job.id,
+            title: job.task.clone(),
+            description: String::new(),
+            state: ui_state.to_string(),
+            user_id: job.user_id.clone(),
+            created_at: job.created_at.to_rfc3339(),
+            started_at: job.started_at.map(|dt| dt.to_rfc3339()),
+            completed_at: job.completed_at.map(|dt| dt.to_rfc3339()),
+            elapsed_secs,
+            project_dir: Some(job.project_dir.clone()),
+            browse_url: Some(format!("/projects/{}/", browse_id)),
+            job_mode: {
+                let mode = store.get_sandbox_job_mode(job.id).await.ok().flatten();
+                mode.filter(|m| m != "worker")
+            },
+            transitions,
+        }));
     }
 
     Err((StatusCode::NOT_FOUND, "Job not found".to_string()))
@@ -1031,31 +1197,36 @@ async fn jobs_cancel_handler(
     let job_id = Uuid::parse_str(&id)
         .map_err(|_| (StatusCode::BAD_REQUEST, "Invalid job ID".to_string()))?;
 
-    // Try sandbox job cancellation.
-    if let Some(ref store) = state.store {
-        if let Ok(Some(job)) = store.get_sandbox_job(job_id).await {
-            if job.status == "running" || job.status == "creating" {
-                // Stop the container if we have a job manager.
-                if let Some(ref jm) = state.job_manager {
-                    let _ = jm.stop_job(job_id).await;
-                }
-                store
-                    .update_sandbox_job_status(
-                        job_id,
-                        "failed",
-                        Some(false),
-                        Some("Cancelled by user"),
-                        None,
-                        Some(chrono::Utc::now()),
-                    )
-                    .await
-                    .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
-            }
-            return Ok(Json(serde_json::json!({
-                "status": "cancelled",
-                "job_id": job_id,
-            })));
+    // Try sandbox job cancellation, scoped to the authenticated user.
+    if let Some(ref store) = state.store
+        && let Ok(Some(job)) = store.get_sandbox_job(job_id).await
+    {
+        if job.user_id != state.user_id {
+            return Err((StatusCode::NOT_FOUND, "Job not found".to_string()));
         }
+        if job.status == "running" || job.status == "creating" {
+            // Stop the container if we have a job manager.
+            if let Some(ref jm) = state.job_manager
+                && let Err(e) = jm.stop_job(job_id).await
+            {
+                tracing::warn!(job_id = %job_id, error = %e, "Failed to stop container during cancellation");
+            }
+            store
+                .update_sandbox_job_status(
+                    job_id,
+                    "failed",
+                    Some(false),
+                    Some("Cancelled by user"),
+                    None,
+                    Some(chrono::Utc::now()),
+                )
+                .await
+                .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
+        }
+        return Ok(Json(serde_json::json!({
+            "status": "cancelled",
+            "job_id": job_id,
+        })));
     }
 
     Err((StatusCode::NOT_FOUND, "Job not found".to_string()))
@@ -1083,6 +1254,11 @@ async fn jobs_restart_handler(
         .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?
         .ok_or((StatusCode::NOT_FOUND, "Job not found".to_string()))?;
 
+    // Scope to the authenticated user.
+    if old_job.user_id != state.user_id {
+        return Err((StatusCode::NOT_FOUND, "Job not found".to_string()));
+    }
+
     if old_job.status != "interrupted" && old_job.status != "failed" {
         return Err((
             StatusCode::CONFLICT,
@@ -1157,6 +1333,16 @@ async fn jobs_prompt_handler(
         .parse()
         .map_err(|_| (StatusCode::BAD_REQUEST, "Invalid job ID".to_string()))?;
 
+    // Verify user owns this job.
+    if let Some(ref store) = state.store
+        && !store
+            .sandbox_job_belongs_to_user(job_id, &state.user_id)
+            .await
+            .unwrap_or(false)
+    {
+        return Err((StatusCode::NOT_FOUND, "Job not found".to_string()));
+    }
+
     let content = body
         .get("content")
         .and_then(|v| v.as_str())
@@ -1195,6 +1381,15 @@ async fn jobs_events_handler(
         .parse()
         .map_err(|_| (StatusCode::BAD_REQUEST, "Invalid job ID".to_string()))?;
 
+    // Verify user owns this job.
+    if !store
+        .sandbox_job_belongs_to_user(job_id, &state.user_id)
+        .await
+        .unwrap_or(false)
+    {
+        return Err((StatusCode::NOT_FOUND, "Job not found".to_string()));
+    }
+
     let events = store
         .list_job_events(job_id)
         .await
@@ -1244,6 +1439,11 @@ async fn job_files_list_handler(
         .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?
         .ok_or((StatusCode::NOT_FOUND, "Job not found".to_string()))?;
 
+    // Verify user owns this job.
+    if job.user_id != state.user_id {
+        return Err((StatusCode::NOT_FOUND, "Job not found".to_string()));
+    }
+
     let base = std::path::PathBuf::from(&job.project_dir);
     let rel_path = query.path.as_deref().unwrap_or("");
     let target = base.join(rel_path);
@@ -1307,6 +1507,11 @@ async fn job_files_read_handler(
         .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?
         .ok_or((StatusCode::NOT_FOUND, "Job not found".to_string()))?;
 
+    // Verify user owns this job.
+    if job.user_id != state.user_id {
+        return Err((StatusCode::NOT_FOUND, "Job not found".to_string()));
+    }
+
     let path = query.path.as_deref().ok_or((
         StatusCode::BAD_REQUEST,
         "path parameter required".to_string(),
@@ -1525,6 +1730,15 @@ async fn project_file_handler(
 /// Shared logic: resolve the file inside `~/.ironclaw/projects/{project_id}/`,
 /// guard against path traversal, and stream the content with the right MIME type.
 async fn serve_project_file(project_id: &str, path: &str) -> axum::response::Response {
+    // Reject project_id values that could escape the projects directory.
+    if project_id.contains('/')
+        || project_id.contains('\\')
+        || project_id.contains("..")
+        || project_id.is_empty()
+    {
+        return (StatusCode::BAD_REQUEST, "Invalid project ID").into_response();
+    }
+
     let base = dirs::home_dir()
         .unwrap_or_else(|| std::path::PathBuf::from("."))
         .join(".ironclaw")

src/channels/web/sse.rs

@@ -13,10 +13,15 @@ use tokio_stream::wrappers::BroadcastStream;
 
 use crate::channels::web::types::SseEvent;
 
+/// Maximum number of concurrent SSE/WebSocket connections.
+/// Prevents resource exhaustion from connection flooding.
+const MAX_CONNECTIONS: u64 = 100;
+
 /// Manages SSE broadcast to all connected browser tabs.
 pub struct SseManager {
     tx: broadcast::Sender<SseEvent>,
     connection_count: Arc<AtomicU64>,
+    max_connections: u64,
 }
 
 impl SseManager {
@@ -27,6 +32,7 @@ impl SseManager {
         Self {
             tx,
             connection_count: Arc::new(AtomicU64::new(0)),
+            max_connections: MAX_CONNECTIONS,
         }
     }
 
@@ -45,25 +51,50 @@ impl SseManager {
     ///
     /// Returns a stream of `SseEvent` values and increments/decrements the
     /// connection counter on creation/drop, just like `subscribe()` does for SSE.
-    pub fn subscribe_raw(&self) -> impl Stream<Item = SseEvent> + Send + 'static + use<> {
+    ///
+    /// Returns `None` if the maximum connection limit has been reached.
+    pub fn subscribe_raw(&self) -> Option<impl Stream<Item = SseEvent> + Send + 'static + use<>> {
+        // Atomically increment only if below the limit. This prevents
+        // concurrent callers from overshooting max_connections.
         let counter = Arc::clone(&self.connection_count);
-        counter.fetch_add(1, Ordering::Relaxed);
+        let max = self.max_connections;
+        counter
+            .fetch_update(Ordering::Relaxed, Ordering::Relaxed, |current| {
+                if current < max {
+                    Some(current + 1)
+                } else {
+                    None
+                }
+            })
+            .ok()?;
         let rx = self.tx.subscribe();
 
         let stream = BroadcastStream::new(rx).filter_map(|result| result.ok());
 
-        CountedStream {
+        Some(CountedStream {
             inner: stream,
             counter,
-        }
+        })
     }
 
     /// Create a new SSE stream for a client connection.
+    ///
+    /// Returns `None` if the maximum connection limit has been reached.
     pub fn subscribe(
         &self,
-    ) -> Sse<impl Stream<Item = Result<Event, Infallible>> + Send + 'static + use<>> {
+    ) -> Option<Sse<impl Stream<Item = Result<Event, Infallible>> + Send + 'static + use<>>> {
+        // Atomically increment only if below the limit.
         let counter = Arc::clone(&self.connection_count);
-        counter.fetch_add(1, Ordering::Relaxed);
+        let max = self.max_connections;
+        counter
+            .fetch_update(Ordering::Relaxed, Ordering::Relaxed, |current| {
+                if current < max {
+                    Some(current + 1)
+                } else {
+                    None
+                }
+            })
+            .ok()?;
         let rx = self.tx.subscribe();
 
         let stream = BroadcastStream::new(rx)
@@ -99,8 +130,10 @@ impl SseManager {
             counter,
         };
 
-        Sse::new(counted_stream)
-            .keep_alive(KeepAlive::new().interval(Duration::from_secs(30)).text(""))
+        Some(
+            Sse::new(counted_stream)
+                .keep_alive(KeepAlive::new().interval(Duration::from_secs(30)).text("")),
+        )
     }
 }
 
@@ -175,7 +208,7 @@ mod tests {
     #[tokio::test]
     async fn test_subscribe_raw_receives_events() {
         let manager = SseManager::new();
-        let mut stream = Box::pin(manager.subscribe_raw());
+        let mut stream = Box::pin(manager.subscribe_raw().expect("should subscribe"));
 
         assert_eq!(manager.connection_count(), 1);
 
@@ -195,7 +228,7 @@ mod tests {
     async fn test_subscribe_raw_decrements_on_drop() {
         let manager = SseManager::new();
         {
-            let _stream = Box::pin(manager.subscribe_raw());
+            let _stream = Box::pin(manager.subscribe_raw().expect("should subscribe"));
             assert_eq!(manager.connection_count(), 1);
         }
         // Stream dropped, counter should decrement
@@ -205,8 +238,8 @@ mod tests {
     #[tokio::test]
     async fn test_subscribe_raw_multiple_subscribers() {
         let manager = SseManager::new();
-        let mut s1 = Box::pin(manager.subscribe_raw());
-        let mut s2 = Box::pin(manager.subscribe_raw());
+        let mut s1 = Box::pin(manager.subscribe_raw().expect("should subscribe"));
+        let mut s2 = Box::pin(manager.subscribe_raw().expect("should subscribe"));
         assert_eq!(manager.connection_count(), 2);
 
         manager.broadcast(SseEvent::Heartbeat);
@@ -221,4 +254,18 @@ mod tests {
         drop(s2);
         assert_eq!(manager.connection_count(), 0);
     }
+
+    #[tokio::test]
+    async fn test_subscribe_raw_rejects_over_limit() {
+        let mut manager = SseManager::new();
+        manager.max_connections = 2; // Low limit for testing
+
+        let _s1 = Box::pin(manager.subscribe_raw().expect("first should succeed"));
+        let _s2 = Box::pin(manager.subscribe_raw().expect("second should succeed"));
+        assert_eq!(manager.connection_count(), 2);
+
+        // Third should be rejected
+        assert!(manager.subscribe_raw().is_none());
+        assert!(manager.subscribe().is_none());
+    }
 }

src/channels/web/static/app.js

@@ -12,6 +12,7 @@ let loadingOlder = false;
 let jobEvents = new Map(); // job_id -> Array of events
 let jobListRefreshTimer = null;
 const JOB_EVENTS_CAP = 500;
+const MEMORY_SEARCH_QUERY_MAX_LENGTH = 100;
 
 // --- Auth ---
 
@@ -281,6 +282,8 @@ function sendApprovalAction(requestId, action) {
 function renderMarkdown(text) {
   if (typeof marked !== 'undefined') {
     let html = marked.parse(text);
+    // Sanitize HTML output to prevent XSS from tool output or LLM responses.
+    html = sanitizeRenderedHtml(html);
     // Inject copy buttons into <pre> blocks
     html = html.replace(/<pre>/g, '<pre class="code-block-wrapper"><button class="copy-btn" onclick="copyCodeBlock(this)">Copy</button>');
     return html;
@@ -288,6 +291,28 @@ function renderMarkdown(text) {
   return escapeHtml(text);
 }
 
+// Strip dangerous HTML elements and attributes from rendered markdown.
+// This prevents XSS from tool output or prompt injection in LLM responses.
+function sanitizeRenderedHtml(html) {
+  html = html.replace(/<script\b[^<]*(?:(?!<\/script>)<[^<]*)*<\/script>/gi, '');
+  html = html.replace(/<iframe\b[^>]*>[\s\S]*?<\/iframe>/gi, '');
+  html = html.replace(/<object\b[^>]*>[\s\S]*?<\/object>/gi, '');
+  html = html.replace(/<embed\b[^>]*\/?>/gi, '');
+  html = html.replace(/<form\b[^>]*>[\s\S]*?<\/form>/gi, '');
+  html = html.replace(/<style\b[^>]*>[\s\S]*?<\/style>/gi, '');
+  html = html.replace(/<link\b[^>]*\/?>/gi, '');
+  html = html.replace(/<base\b[^>]*\/?>/gi, '');
+  html = html.replace(/<meta\b[^>]*\/?>/gi, '');
+  // Remove event handler attributes (onclick, onerror, onload, etc.)
+  html = html.replace(/\s+on\w+\s*=\s*"[^"]*"/gi, '');
+  html = html.replace(/\s+on\w+\s*=\s*'[^']*'/gi, '');
+  html = html.replace(/\s+on\w+\s*=\s*[^\s>]+/gi, '');
+  // Remove javascript: and data: URLs in href/src attributes
+  html = html.replace(/(href|src|action)\s*=\s*["']?\s*javascript\s*:/gi, '$1="');
+  html = html.replace(/(href|src|action)\s*=\s*["']?\s*data\s*:/gi, '$1="');
+  return html;
+}
+
 function copyCodeBlock(btn) {
   const pre = btn.parentElement;
   const code = pre.querySelector('code');
@@ -977,9 +1002,12 @@ function buildBreadcrumb(path) {
 }
 
 function searchMemory(query) {
+  const normalizedQuery = normalizeSearchQuery(query);
+  if (!normalizedQuery) return;
+
   apiFetch('/api/memory/search', {
     method: 'POST',
-    body: { query, limit: 20 },
+    body: { query: normalizedQuery, limit: 20 },
   }).then((data) => {
     const tree = document.getElementById('memory-tree');
     tree.innerHTML = '';
@@ -990,18 +1018,23 @@ function searchMemory(query) {
     for (const result of data.results) {
       const item = document.createElement('div');
       item.className = 'search-result';
-      const snippet = snippetAround(result.content, query, 120);
+      const snippet = snippetAround(result.content, normalizedQuery, 120);
       item.innerHTML = '<div class="path">' + escapeHtml(result.path) + '</div>'
-        + '<div class="snippet">' + highlightQuery(snippet, query) + '</div>';
+        + '<div class="snippet">' + highlightQuery(snippet, normalizedQuery) + '</div>';
       item.addEventListener('click', () => readMemoryFile(result.path));
       tree.appendChild(item);
     }
   }).catch(() => {});
 }
 
+function normalizeSearchQuery(query) {
+  return (typeof query === 'string' ? query : '').slice(0, MEMORY_SEARCH_QUERY_MAX_LENGTH);
+}
+
 function snippetAround(text, query, len) {
+  const normalizedQuery = normalizeSearchQuery(query);
   const lower = text.toLowerCase();
-  const idx = lower.indexOf(query.toLowerCase());
+  const idx = lower.indexOf(normalizedQuery.toLowerCase());
   if (idx < 0) return text.substring(0, len);
   const start = Math.max(0, idx - Math.floor(len / 2));
   const end = Math.min(text.length, start + len);
@@ -1014,11 +1047,11 @@ function snippetAround(text, query, len) {
 function highlightQuery(text, query) {
   if (!query) return escapeHtml(text);
   const escaped = escapeHtml(text);
-  const queryEscaped = query.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+  const normalizedQuery = normalizeSearchQuery(query);
+  const queryEscaped = normalizedQuery.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
   const re = new RegExp('(' + queryEscaped + ')', 'gi');
   return escaped.replace(re, '<mark>$1</mark>');
 }
-
 // --- Logs ---
 
 const LOG_MAX_ENTRIES = 2000;

src/channels/web/static/index.html

@@ -5,7 +5,11 @@
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <title>IronClaw</title>
   <link rel="stylesheet" href="/style.css">
-  <script src="https://cdn.jsdelivr.net/npm/marked/marked.min.js"></script>
+  <script
+    src="https://cdn.jsdelivr.net/npm/[email protected]/lib/marked.umd.min.js"
+    integrity="sha384-pN9zSKOnTZwXRtYZAu0PBPEgR2B7DOC1aeLxQ33oJ0oy5iN1we6gm57xldM2irDG"
+    crossorigin="anonymous"
+  ></script>
 </head>
 <body>
   <!-- Auth Screen -->

src/channels/web/ws.rs

@@ -71,8 +71,17 @@ pub async fn handle_ws_connection(socket: WebSocket, state: Arc<GatewayState>) {
     }
     let tracker_for_drop = state.ws_tracker.clone();
 
-    // Subscribe to broadcast events (same source as SSE)
-    let mut event_stream = Box::pin(state.sse.subscribe_raw());
+    // Subscribe to broadcast events (same source as SSE).
+    // Reject if we've hit the connection limit.
+    let Some(raw_stream) = state.sse.subscribe_raw() else {
+        tracing::warn!("WebSocket rejected: too many connections");
+        // Decrement the WS tracker we already incremented above.
+        if let Some(ref tracker) = tracker_for_drop {
+            tracker.decrement();
+        }
+        return;
+    };
+    let mut event_stream = Box::pin(raw_stream);
 
     // Channel for the sender task to receive messages from both
     // the broadcast stream and any direct sends (like Pong)
@@ -476,6 +485,8 @@ mod tests {
             user_id: "test".to_string(),
             shutdown_tx: tokio::sync::RwLock::new(None),
             ws_tracker: Some(Arc::new(WsConnectionTracker::new())),
+            llm_provider: None,
+            chat_rate_limiter: crate::channels::web::server::RateLimiter::new(30, 60),
         }
     }
 }

src/cli/config.rs

@@ -1,7 +1,9 @@
 //! Configuration management CLI commands.
 //!
 //! Commands for viewing and modifying settings.
-//! Settings are stored in PostgreSQL (env > DB > default).
+//! Settings are stored in the database (env > DB > default).
+
+use std::sync::Arc;
 
 use clap::Subcommand;
 
@@ -46,11 +48,9 @@ pub enum ConfigCommand {
 /// Connects to the database to read/write settings. Falls back to disk
 /// if the database is not available.
 pub async fn run_config_command(cmd: ConfigCommand) -> anyhow::Result<()> {
-    let _ = dotenvy::dotenv();
-
     // Try to connect to the DB for settings access
-    let store = match connect_store().await {
-        Ok(s) => Some(s),
+    let db: Option<Arc<dyn crate::db::Database>> = match connect_db().await {
+        Ok(d) => Some(d),
         Err(e) => {
             eprintln!(
                 "Warning: Could not connect to database ({}), using disk fallback",
@@ -60,41 +60,42 @@ pub async fn run_config_command(cmd: ConfigCommand) -> anyhow::Result<()> {
         }
     };
 
+    let db_ref = db.as_deref();
     match cmd {
-        ConfigCommand::List { filter } => list_settings(store.as_ref(), filter).await,
-        ConfigCommand::Get { path } => get_setting(store.as_ref(), &path).await,
-        ConfigCommand::Set { path, value } => set_setting(store.as_ref(), &path, &value).await,
-        ConfigCommand::Reset { path } => reset_setting(store.as_ref(), &path).await,
-        ConfigCommand::Path => show_path(store.is_some()),
+        ConfigCommand::List { filter } => list_settings(db_ref, filter).await,
+        ConfigCommand::Get { path } => get_setting(db_ref, &path).await,
+        ConfigCommand::Set { path, value } => set_setting(db_ref, &path, &value).await,
+        ConfigCommand::Reset { path } => reset_setting(db_ref, &path).await,
+        ConfigCommand::Path => show_path(db_ref.is_some()),
     }
 }
 
-/// Bootstrap a DB connection for config commands.
-async fn connect_store() -> anyhow::Result<crate::history::Store> {
+/// Bootstrap a DB connection for config commands (backend-agnostic).
+async fn connect_db() -> anyhow::Result<Arc<dyn crate::db::Database>> {
     let config = crate::config::Config::from_env()
         .await
         .map_err(|e| anyhow::anyhow!("{}", e))?;
-    let store = crate::history::Store::new(&config.database).await?;
-    store.run_migrations().await?;
-    Ok(store)
+    crate::db::connect_from_config(&config.database)
+        .await
+        .map_err(|e| anyhow::anyhow!("{}", e))
 }
 
 const DEFAULT_USER_ID: &str = "default";
 
 /// Load settings: DB if available, else disk.
-async fn load_settings(store: Option<&crate::history::Store>) -> Settings {
+async fn load_settings(store: Option<&dyn crate::db::Database>) -> Settings {
     if let Some(store) = store {
         match store.get_all_settings(DEFAULT_USER_ID).await {
             Ok(map) if !map.is_empty() => return Settings::from_db_map(&map),
             _ => {}
         }
     }
-    Settings::load()
+    Settings::default()
 }
 
 /// List all settings.
 async fn list_settings(
-    store: Option<&crate::history::Store>,
+    store: Option<&dyn crate::db::Database>,
     filter: Option<String>,
 ) -> anyhow::Result<()> {
     let settings = load_settings(store).await;
@@ -107,10 +108,10 @@ async fn list_settings(
     println!();
 
     for (key, value) in all {
-        if let Some(ref f) = filter {
-            if !key.starts_with(f) {
-                continue;
-            }
+        if let Some(ref f) = filter
+            && !key.starts_with(f)
+        {
+            continue;
         }
 
         let display_value = if value.len() > 60 {
@@ -126,7 +127,7 @@ async fn list_settings(
 }
 
 /// Get a specific setting.
-async fn get_setting(store: Option<&crate::history::Store>, path: &str) -> anyhow::Result<()> {
+async fn get_setting(store: Option<&dyn crate::db::Database>, path: &str) -> anyhow::Result<()> {
     let settings = load_settings(store).await;
 
     match settings.get(path) {
@@ -142,7 +143,7 @@ async fn get_setting(store: Option<&crate::history::Store>, path: &str) -> anyho
 
 /// Set a setting value.
 async fn set_setting(
-    store: Option<&crate::history::Store>,
+    store: Option<&dyn crate::db::Database>,
     path: &str,
     value: &str,
 ) -> anyhow::Result<()> {
@@ -152,42 +153,36 @@ async fn set_setting(
         .set(path, value)
         .map_err(|e| anyhow::anyhow!("{}", e))?;
 
-    // Save to DB if available, otherwise disk
-    if let Some(store) = store {
-        let json_value = match serde_json::from_str::<serde_json::Value>(value) {
-            Ok(v) => v,
-            Err(_) => serde_json::Value::String(value.to_string()),
-        };
-        store
-            .set_setting(DEFAULT_USER_ID, path, &json_value)
-            .await
-            .map_err(|e| anyhow::anyhow!("Failed to save to database: {}", e))?;
-    } else {
-        settings.save()?;
-    }
+    let store = store.ok_or_else(|| {
+        anyhow::anyhow!("Database connection required to save settings. Check DATABASE_URL.")
+    })?;
+    let json_value = match serde_json::from_str::<serde_json::Value>(value) {
+        Ok(v) => v,
+        Err(_) => serde_json::Value::String(value.to_string()),
+    };
+    store
+        .set_setting(DEFAULT_USER_ID, path, &json_value)
+        .await
+        .map_err(|e| anyhow::anyhow!("Failed to save to database: {}", e))?;
 
     println!("Set {} = {}", path, value);
     Ok(())
 }
 
 /// Reset a setting to default.
-async fn reset_setting(store: Option<&crate::history::Store>, path: &str) -> anyhow::Result<()> {
+async fn reset_setting(store: Option<&dyn crate::db::Database>, path: &str) -> anyhow::Result<()> {
     let default = Settings::default();
     let default_value = default
         .get(path)
         .ok_or_else(|| anyhow::anyhow!("Unknown setting: {}", path))?;
 
-    // Delete from DB (falling back to default) or reset on disk
-    if let Some(store) = store {
-        store
-            .delete_setting(DEFAULT_USER_ID, path)
-            .await
-            .map_err(|e| anyhow::anyhow!("Failed to delete setting from database: {}", e))?;
-    } else {
-        let mut settings = Settings::load();
-        settings.reset(path).map_err(|e| anyhow::anyhow!("{}", e))?;
-        settings.save()?;
-    }
+    let store = store.ok_or_else(|| {
+        anyhow::anyhow!("Database connection required to reset settings. Check DATABASE_URL.")
+    })?;
+    store
+        .delete_setting(DEFAULT_USER_ID, path)
+        .await
+        .map_err(|e| anyhow::anyhow!("Failed to delete setting from database: {}", e))?;
 
     println!("Reset {} to default: {}", path, default_value);
     Ok(())
@@ -196,38 +191,14 @@ async fn reset_setting(store: Option<&crate::history::Store>, path: &str) -> any
 /// Show the settings storage info.
 fn show_path(has_db: bool) -> anyhow::Result<()> {
     if has_db {
-        println!("Settings stored in: PostgreSQL (settings table)");
-        println!(
-            "Bootstrap config:   {}",
-            crate::bootstrap::BootstrapConfig::default_path().display()
-        );
+        println!("Settings stored in: database (settings table)");
     } else {
-        let path = Settings::default_path();
-        println!("Settings stored in: {} (disk fallback)", path.display());
-
-        if path.exists() {
-            let metadata = std::fs::metadata(&path)?;
-            println!("  Size: {} bytes", metadata.len());
-            if let Ok(modified) = metadata.modified() {
-                use std::time::SystemTime;
-                let duration = SystemTime::now()
-                    .duration_since(modified)
-                    .unwrap_or_default();
-                let secs = duration.as_secs();
-                if secs < 60 {
-                    println!("  Modified: {} seconds ago", secs);
-                } else if secs < 3600 {
-                    println!("  Modified: {} minutes ago", secs / 60);
-                } else if secs < 86400 {
-                    println!("  Modified: {} hours ago", secs / 3600);
-                } else {
-                    println!("  Modified: {} days ago", secs / 86400);
-                }
-            }
-        } else {
-            println!("  (does not exist, using defaults)");
-        }
+        println!("Settings stored in: PostgreSQL (not connected, using defaults)");
     }
+    println!(
+        "Env config:         {}",
+        crate::bootstrap::ironclaw_env_path().display()
+    );
 
     Ok(())
 }

src/cli/mcp.rs

@@ -8,8 +8,10 @@ use std::sync::Arc;
 use clap::Subcommand;
 
 use crate::config::Config;
-use crate::history::Store;
-use crate::secrets::{PostgresSecretsStore, SecretsCrypto, SecretsStore};
+use crate::db::Database;
+#[cfg(feature = "postgres")]
+use crate::secrets::PostgresSecretsStore;
+use crate::secrets::{SecretsCrypto, SecretsStore};
 use crate::tools::mcp::{
     McpClient, McpServerConfig, McpSessionManager, OAuthConfig,
     auth::{authorize_mcp_server, is_authenticated},
@@ -172,10 +174,10 @@ async fn add_server(
     config.validate()?;
 
     // Save (DB if available, else disk)
-    let store = connect_store().await;
-    let mut servers = load_servers(store.as_ref()).await?;
+    let db = connect_db().await;
+    let mut servers = load_servers(db.as_deref()).await?;
     servers.upsert(config);
-    save_servers(store.as_ref(), &servers).await?;
+    save_servers(db.as_deref(), &servers).await?;
 
     println!();
     println!("  ✓ Added MCP server '{}'", name);
@@ -193,12 +195,12 @@ async fn add_server(
 
 /// Remove an MCP server.
 async fn remove_server(name: String) -> anyhow::Result<()> {
-    let store = connect_store().await;
-    let mut servers = load_servers(store.as_ref()).await?;
+    let db = connect_db().await;
+    let mut servers = load_servers(db.as_deref()).await?;
     if !servers.remove(&name) {
         anyhow::bail!("Server '{}' not found", name);
     }
-    save_servers(store.as_ref(), &servers).await?;
+    save_servers(db.as_deref(), &servers).await?;
 
     println!();
     println!("  ✓ Removed MCP server '{}'", name);
@@ -209,8 +211,8 @@ async fn remove_server(name: String) -> anyhow::Result<()> {
 
 /// List configured MCP servers.
 async fn list_servers(verbose: bool) -> anyhow::Result<()> {
-    let store = connect_store().await;
-    let servers = load_servers(store.as_ref()).await?;
+    let db = connect_db().await;
+    let servers = load_servers(db.as_deref()).await?;
 
     if servers.servers.is_empty() {
         println!();
@@ -268,8 +270,8 @@ async fn list_servers(verbose: bool) -> anyhow::Result<()> {
 /// Authenticate with an MCP server.
 async fn auth_server(name: String, user_id: String) -> anyhow::Result<()> {
     // Get server config
-    let store = connect_store().await;
-    let servers = load_servers(store.as_ref()).await?;
+    let db = connect_db().await;
+    let servers = load_servers(db.as_deref()).await?;
     let server = servers
         .get(&name)
         .cloned()
@@ -341,8 +343,8 @@ async fn auth_server(name: String, user_id: String) -> anyhow::Result<()> {
 /// Test connection to an MCP server.
 async fn test_server(name: String, user_id: String) -> anyhow::Result<()> {
     // Get server config
-    let store = connect_store().await;
-    let servers = load_servers(store.as_ref()).await?;
+    let db = connect_db().await;
+    let servers = load_servers(db.as_deref()).await?;
     let server = servers
         .get(&name)
         .cloned()
@@ -437,8 +439,8 @@ async fn test_server(name: String, user_id: String) -> anyhow::Result<()> {
 
 /// Toggle server enabled/disabled state.
 async fn toggle_server(name: String, enable: bool, disable: bool) -> anyhow::Result<()> {
-    let store = connect_store().await;
-    let mut servers = load_servers(store.as_ref()).await?;
+    let db = connect_db().await;
+    let mut servers = load_servers(db.as_deref()).await?;
 
     let server = servers
         .get_mut(&name)
@@ -453,7 +455,7 @@ async fn toggle_server(name: String, enable: bool, disable: bool) -> anyhow::Res
     };
 
     server.enabled = new_state;
-    save_servers(store.as_ref(), &servers).await?;
+    save_servers(db.as_deref(), &servers).await?;
 
     let status = if new_state { "enabled" } else { "disabled" };
     println!();
@@ -465,18 +467,16 @@ async fn toggle_server(name: String, enable: bool, disable: bool) -> anyhow::Res
 
 const DEFAULT_USER_ID: &str = "default";
 
-/// Try to connect to the database store for DB-backed config.
-async fn connect_store() -> Option<Store> {
+/// Try to connect to the database (backend-agnostic).
+async fn connect_db() -> Option<Arc<dyn Database>> {
     let config = Config::from_env().await.ok()?;
-    let store = Store::new(&config.database).await.ok()?;
-    store.run_migrations().await.ok()?;
-    Some(store)
+    crate::db::connect_from_config(&config.database).await.ok()
 }
 
 /// Load MCP servers (DB if available, else disk).
-async fn load_servers(store: Option<&Store>) -> Result<McpServersFile, config::ConfigError> {
-    if let Some(store) = store {
-        config::load_mcp_servers_from_db(store, DEFAULT_USER_ID).await
+async fn load_servers(db: Option<&dyn Database>) -> Result<McpServersFile, config::ConfigError> {
+    if let Some(db) = db {
+        config::load_mcp_servers_from_db(db, DEFAULT_USER_ID).await
     } else {
         config::load_mcp_servers().await
     }
@@ -484,11 +484,11 @@ async fn load_servers(store: Option<&Store>) -> Result<McpServersFile, config::C
 
 /// Save MCP servers (DB if available, else disk).
 async fn save_servers(
-    store: Option<&Store>,
+    db: Option<&dyn Database>,
     servers: &McpServersFile,
 ) -> Result<(), config::ConfigError> {
-    if let Some(store) = store {
-        config::save_mcp_servers_to_db(store, DEFAULT_USER_ID, servers).await
+    if let Some(db) = db {
+        config::save_mcp_servers_to_db(db, DEFAULT_USER_ID, servers).await
     } else {
         config::save_mcp_servers(servers).await
     }
@@ -504,14 +504,61 @@ async fn get_secrets_store() -> anyhow::Result<Arc<dyn SecretsStore + Send + Syn
         )
     })?;
 
-    let store = Store::new(&config.database).await?;
-    store.run_migrations().await?;
-
     let crypto = SecretsCrypto::new(master_key.clone())?;
-    Ok(Arc::new(PostgresSecretsStore::new(
-        store.pool(),
-        Arc::new(crypto),
-    )))
+
+    #[cfg(feature = "postgres")]
+    {
+        let store = crate::history::Store::new(&config.database).await?;
+        store.run_migrations().await?;
+        Ok(Arc::new(PostgresSecretsStore::new(
+            store.pool(),
+            Arc::new(crypto),
+        )))
+    }
+
+    #[cfg(all(feature = "libsql", not(feature = "postgres")))]
+    {
+        use crate::db::Database as _;
+        use crate::db::libsql_backend::LibSqlBackend;
+        use secrecy::ExposeSecret as _;
+
+        let default_path = crate::config::default_libsql_path();
+        let db_path = config
+            .database
+            .libsql_path
+            .as_deref()
+            .unwrap_or(&default_path);
+
+        let backend = if let Some(ref url) = config.database.libsql_url {
+            let token = config.database.libsql_auth_token.as_ref().ok_or_else(|| {
+                anyhow::anyhow!("LIBSQL_AUTH_TOKEN is required when LIBSQL_URL is set")
+            })?;
+            LibSqlBackend::new_remote_replica(db_path, url, token.expose_secret())
+                .await
+                .map_err(|e| anyhow::anyhow!("{}", e))?
+        } else {
+            LibSqlBackend::new_local(db_path)
+                .await
+                .map_err(|e| anyhow::anyhow!("{}", e))?
+        };
+        backend
+            .run_migrations()
+            .await
+            .map_err(|e| anyhow::anyhow!("{}", e))?;
+
+        return Ok(Arc::new(crate::secrets::LibSqlSecretsStore::new(
+            backend.shared_db(),
+            Arc::new(crypto),
+        )));
+    }
+
+    #[cfg(not(any(feature = "postgres", feature = "libsql")))]
+    {
+        let _ = crypto;
+        anyhow::bail!(
+            "No database backend available for secrets. Enable 'postgres' or 'libsql' feature."
+        );
+    }
 }
 
 #[cfg(test)]

src/cli/memory.rs

@@ -9,6 +9,30 @@ use clap::Subcommand;
 
 use crate::workspace::{EmbeddingProvider, SearchConfig, Workspace};
 
+/// Run a memory command using the Database trait (works with any backend).
+pub async fn run_memory_command_with_db(
+    cmd: MemoryCommand,
+    db: std::sync::Arc<dyn crate::db::Database>,
+    embeddings: Option<Arc<dyn EmbeddingProvider>>,
+) -> anyhow::Result<()> {
+    let mut workspace = Workspace::new_with_db("default", db);
+    if let Some(emb) = embeddings {
+        workspace = workspace.with_embeddings(emb);
+    }
+
+    match cmd {
+        MemoryCommand::Search { query, limit } => search(&workspace, &query, limit).await,
+        MemoryCommand::Read { path } => read(&workspace, &path).await,
+        MemoryCommand::Write {
+            path,
+            content,
+            append,
+        } => write(&workspace, &path, content, append).await,
+        MemoryCommand::Tree { path, depth } => tree(&workspace, &path, depth).await,
+        MemoryCommand::Status => status(&workspace).await,
+    }
+}
+
 #[derive(Subcommand, Debug, Clone)]
 pub enum MemoryCommand {
     /// Search workspace memory (hybrid full-text + semantic)
@@ -55,7 +79,8 @@ pub enum MemoryCommand {
     Status,
 }
 
-/// Run a memory command.
+/// Run a memory command (PostgreSQL backend).
+#[cfg(feature = "postgres")]
 pub async fn run_memory_command(
     cmd: MemoryCommand,
     pool: deadpool_postgres::Pool,

src/cli/mod.rs

@@ -12,13 +12,17 @@
 mod config;
 mod mcp;
 pub mod memory;
+pub mod oauth_defaults;
 mod pairing;
 pub mod status;
 mod tool;
 
 pub use config::{ConfigCommand, run_config_command};
 pub use mcp::{McpCommand, run_mcp_command};
-pub use memory::{MemoryCommand, run_memory_command};
+pub use memory::MemoryCommand;
+#[cfg(feature = "postgres")]
+pub use memory::run_memory_command;
+pub use memory::run_memory_command_with_db;
 pub use pairing::{PairingCommand, run_pairing_command, run_pairing_command_with_store};
 pub use status::run_status_command;
 pub use tool::{ToolCommand, run_tool_command};

src/cli/oauth_defaults.rs

@@ -0,0 +1,342 @@
+//! Shared OAuth infrastructure: built-in credentials, callback server, landing pages.
+//!
+//! Every OAuth flow in the codebase (WASM tool auth, MCP server auth, NEAR AI login)
+//! uses the same callback port, landing page, and listener logic from this module.
+//!
+//! # Built-in Credentials
+//!
+//! Many CLI tools (gcloud, rclone, gdrive) ship with default OAuth credentials
+//! so users don't need to register their own OAuth app. Google explicitly
+//! documents that client_secret for "Desktop App" / "Installed App" types
+//! is NOT actually secret.
+//!
+//! Default credentials are hardcoded below. They can be overridden at:
+//!
+//! - **Compile time**: Set IRONCLAW_GOOGLE_CLIENT_ID / IRONCLAW_GOOGLE_CLIENT_SECRET
+//!   env vars before building to replace the hardcoded defaults.
+//! - **Runtime**: Users can set GOOGLE_OAUTH_CLIENT_ID / GOOGLE_OAUTH_CLIENT_SECRET
+//!   env vars, which take priority over built-in defaults.
+
+use std::time::Duration;
+
+use tokio::io::{AsyncBufReadExt, AsyncWriteExt, BufReader};
+use tokio::net::TcpListener;
+
+// ── Built-in credentials ────────────────────────────────────────────────
+
+pub struct OAuthCredentials {
+    pub client_id: &'static str,
+    pub client_secret: &'static str,
+}
+
+/// Google OAuth "Desktop App" credentials, shared across all Google tools.
+/// Compile-time env vars override the hardcoded defaults below.
+const GOOGLE_CLIENT_ID: &str = match option_env!("IRONCLAW_GOOGLE_CLIENT_ID") {
+    Some(v) => v,
+    None => "564604149681-efo25d43rs85v0tibdepsmdv5dsrhhr0.apps.googleusercontent.com",
+};
+const GOOGLE_CLIENT_SECRET: &str = match option_env!("IRONCLAW_GOOGLE_CLIENT_SECRET") {
+    Some(v) => v,
+    None => "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",
+};
+
+/// Returns built-in OAuth credentials for a provider, keyed by secret_name.
+///
+/// The secret_name comes from the tool's capabilities.json `auth.secret_name` field.
+/// Returns `None` if no built-in credentials are configured for that provider.
+pub fn builtin_credentials(secret_name: &str) -> Option<OAuthCredentials> {
+    match secret_name {
+        "google_oauth_token" => Some(OAuthCredentials {
+            client_id: GOOGLE_CLIENT_ID,
+            client_secret: GOOGLE_CLIENT_SECRET,
+        }),
+        _ => None,
+    }
+}
+
+// ── Shared callback server ──────────────────────────────────────────────
+
+/// Fixed port for all OAuth callbacks.
+///
+/// Every redirect URI registered with providers must use this port:
+/// `http://localhost:9876/callback` (or `/auth/callback` for NEAR AI).
+pub const OAUTH_CALLBACK_PORT: u16 = 9876;
+
+/// Error from the OAuth callback listener.
+#[derive(Debug, thiserror::Error)]
+pub enum OAuthCallbackError {
+    #[error("Port {0} is in use (another auth flow running?): {1}")]
+    PortInUse(u16, String),
+
+    #[error("Authorization denied by user")]
+    Denied,
+
+    #[error("Timed out waiting for authorization")]
+    Timeout,
+
+    #[error("IO error: {0}")]
+    Io(String),
+}
+
+/// Bind the OAuth callback listener on the fixed port.
+///
+/// Binds to IPv4 `127.0.0.1` first because callback URLs use `127.0.0.1`
+/// explicitly (e.g., NEAR AI redirects to `http://127.0.0.1:9876/auth/callback`).
+/// Falls back to IPv6 `[::1]` only if IPv4 binding fails for a reason other
+/// than `AddrInUse`. If the port is already occupied, fails immediately.
+pub async fn bind_callback_listener() -> Result<TcpListener, OAuthCallbackError> {
+    let ipv4_addr = format!("127.0.0.1:{}", OAUTH_CALLBACK_PORT);
+    match TcpListener::bind(&ipv4_addr).await {
+        Ok(listener) => return Ok(listener),
+        Err(e) if e.kind() == std::io::ErrorKind::AddrInUse => {
+            return Err(OAuthCallbackError::PortInUse(
+                OAUTH_CALLBACK_PORT,
+                e.to_string(),
+            ));
+        }
+        Err(_) => {
+            // IPv4 not available, fall back to IPv6
+        }
+    }
+    TcpListener::bind(format!("[::1]:{}", OAUTH_CALLBACK_PORT))
+        .await
+        .map_err(|e| {
+            if e.kind() == std::io::ErrorKind::AddrInUse {
+                OAuthCallbackError::PortInUse(OAUTH_CALLBACK_PORT, e.to_string())
+            } else {
+                OAuthCallbackError::Io(e.to_string())
+            }
+        })
+}
+
+/// Wait for an OAuth callback and extract a query parameter value.
+///
+/// Listens for a GET request matching `path_prefix` (e.g., "/callback" or "/auth/callback"),
+/// extracts the value of `param_name` (e.g., "code" or "token"), and shows a branded
+/// landing page using `display_name` (e.g., "Google", "Notion", "NEAR AI").
+///
+/// Times out after 5 minutes.
+pub async fn wait_for_callback(
+    listener: TcpListener,
+    path_prefix: &str,
+    param_name: &str,
+    display_name: &str,
+) -> Result<String, OAuthCallbackError> {
+    let path_prefix = path_prefix.to_string();
+    let param_name = param_name.to_string();
+    let display_name = display_name.to_string();
+
+    tokio::time::timeout(Duration::from_secs(300), async move {
+        loop {
+            let (mut socket, _) = listener
+                .accept()
+                .await
+                .map_err(|e| OAuthCallbackError::Io(e.to_string()))?;
+
+            let mut reader = BufReader::new(&mut socket);
+            let mut request_line = String::new();
+            reader
+                .read_line(&mut request_line)
+                .await
+                .map_err(|e| OAuthCallbackError::Io(e.to_string()))?;
+
+            if let Some(path) = request_line.split_whitespace().nth(1)
+                && path.starts_with(&path_prefix)
+                && let Some(query) = path.split('?').nth(1)
+            {
+                // Check for error first
+                if query.contains("error=") {
+                    let html = landing_html(&display_name, false);
+                    let response = format!(
+                        "HTTP/1.1 400 Bad Request\r\n\
+                         Content-Type: text/html; charset=utf-8\r\n\
+                         Connection: close\r\n\
+                         \r\n\
+                         {}",
+                        html
+                    );
+                    let _ = socket.write_all(response.as_bytes()).await;
+                    return Err(OAuthCallbackError::Denied);
+                }
+
+                // Look for the target parameter
+                for param in query.split('&') {
+                    let parts: Vec<&str> = param.splitn(2, '=').collect();
+                    if parts.len() == 2 && parts[0] == param_name {
+                        let value = urlencoding::decode(parts[1])
+                            .unwrap_or_else(|_| parts[1].into())
+                            .into_owned();
+
+                        let html = landing_html(&display_name, true);
+                        let response = format!(
+                            "HTTP/1.1 200 OK\r\n\
+                             Content-Type: text/html; charset=utf-8\r\n\
+                             Connection: close\r\n\
+                             \r\n\
+                             {}",
+                            html
+                        );
+                        let _ = socket.write_all(response.as_bytes()).await;
+                        let _ = socket.shutdown().await;
+
+                        return Ok(value);
+                    }
+                }
+            }
+
+            // Not the callback we're looking for
+            let response = "HTTP/1.1 404 Not Found\r\nConnection: close\r\n\r\n";
+            let _ = socket.write_all(response.as_bytes()).await;
+        }
+    })
+    .await
+    .map_err(|_| OAuthCallbackError::Timeout)?
+}
+
+/// Escape a string for safe interpolation into HTML content.
+fn html_escape(s: &str) -> String {
+    let mut out = String::with_capacity(s.len());
+    for c in s.chars() {
+        match c {
+            '&' => out.push_str("&amp;"),
+            '<' => out.push_str("&lt;"),
+            '>' => out.push_str("&gt;"),
+            '"' => out.push_str("&quot;"),
+            '\'' => out.push_str("&#x27;"),
+            _ => out.push(c),
+        }
+    }
+    out
+}
+
+/// HTML landing page shown in the browser after an OAuth redirect.
+pub fn landing_html(provider_name: &str, success: bool) -> String {
+    let safe_name = html_escape(provider_name);
+    let (icon, heading, subtitle, accent) = if success {
+        (
+            r##"<div style="width:64px;height:64px;border-radius:50%;background:#22c55e;display:flex;align-items:center;justify-content:center;margin:0 auto 24px">
+                <svg width="32" height="32" viewBox="0 0 24 24" fill="none" stroke="#fff" stroke-width="3" stroke-linecap="round" stroke-linejoin="round"><polyline points="20 6 9 17 4 12"/></svg>
+              </div>"##,
+            format!("{} Connected", safe_name),
+            "You can close this window and return to your terminal.",
+            "#22c55e",
+        )
+    } else {
+        (
+            r##"<div style="width:64px;height:64px;border-radius:50%;background:#ef4444;display:flex;align-items:center;justify-content:center;margin:0 auto 24px">
+                <svg width="32" height="32" viewBox="0 0 24 24" fill="none" stroke="#fff" stroke-width="3" stroke-linecap="round" stroke-linejoin="round"><line x1="18" y1="6" x2="6" y2="18"/><line x1="6" y1="6" x2="18" y2="18"/></svg>
+              </div>"##,
+            "Authorization Failed".to_string(),
+            "The request was denied. You can close this window and try again.",
+            "#ef4444",
+        )
+    };
+
+    format!(
+        r#"<!DOCTYPE html>
+<html lang="en">
+<head>
+<meta charset="utf-8">
+<meta name="viewport" content="width=device-width,initial-scale=1">
+<title>IronClaw - {heading}</title>
+<style>
+  * {{ margin:0; padding:0; box-sizing:border-box }}
+  body {{
+    font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, Helvetica, Arial, sans-serif;
+    background: #0a0a0a;
+    color: #e5e5e5;
+    display: flex;
+    justify-content: center;
+    align-items: center;
+    min-height: 100vh;
+  }}
+  .card {{
+    text-align: center;
+    padding: 48px 40px;
+    max-width: 420px;
+    border: 1px solid #262626;
+    border-radius: 16px;
+    background: #141414;
+  }}
+  h1 {{
+    font-size: 22px;
+    font-weight: 600;
+    margin-bottom: 8px;
+    color: #fafafa;
+  }}
+  p {{
+    font-size: 14px;
+    color: #a3a3a3;
+    line-height: 1.5;
+  }}
+  .accent {{ color: {accent}; }}
+  .brand {{
+    margin-top: 32px;
+    font-size: 12px;
+    color: #525252;
+    letter-spacing: 0.5px;
+    text-transform: uppercase;
+  }}
+</style>
+</head>
+<body>
+  <div class="card">
+    {icon}
+    <h1>{heading}</h1>
+    <p>{subtitle}</p>
+    <div class="brand">IronClaw</div>
+  </div>
+</body>
+</html>"#,
+        heading = heading,
+        icon = icon,
+        subtitle = subtitle,
+        accent = accent,
+    )
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::cli::oauth_defaults::{builtin_credentials, landing_html};
+
+    #[test]
+    fn test_unknown_provider_returns_none() {
+        assert!(builtin_credentials("unknown_token").is_none());
+    }
+
+    #[test]
+    fn test_google_returns_based_on_compile_env() {
+        let creds = builtin_credentials("google_oauth_token");
+        assert!(creds.is_some());
+        let creds = creds.unwrap();
+        assert!(!creds.client_id.is_empty());
+        assert!(!creds.client_secret.is_empty());
+    }
+
+    #[test]
+    fn test_landing_html_success_contains_key_elements() {
+        let html = landing_html("Google", true);
+        assert!(html.contains("Google Connected"));
+        assert!(html.contains("charset"));
+        assert!(html.contains("IronClaw"));
+        assert!(html.contains("#22c55e")); // green accent
+        assert!(!html.contains("Failed"));
+    }
+
+    #[test]
+    fn test_landing_html_escapes_provider_name() {
+        let html = landing_html("<script>alert(1)</script>", true);
+        assert!(!html.contains("<script>"));
+        assert!(html.contains("&lt;script&gt;"));
+    }
+
+    #[test]
+    fn test_landing_html_error_contains_key_elements() {
+        let html = landing_html("Notion", false);
+        assert!(html.contains("Authorization Failed"));
+        assert!(html.contains("charset"));
+        assert!(html.contains("IronClaw"));
+        assert!(html.contains("#ef4444")); // red accent
+        assert!(!html.contains("Connected"));
+    }
+}

src/cli/status.rs

@@ -9,7 +9,7 @@ use crate::settings::Settings;
 
 /// Run the status command, printing system health info.
 pub async fn run_status_command() -> anyhow::Result<()> {
-    let settings = Settings::load();
+    let settings = Settings::default();
 
     println!("IronClaw Status");
     println!("===============\n");
@@ -22,16 +22,36 @@ pub async fn run_status_command() -> anyhow::Result<()> {
     );
 
     // Database
-    let db_url_set = settings.database_url.is_some() || std::env::var("DATABASE_URL").is_ok();
     print!("  Database:    ");
-    if db_url_set {
-        // Try to connect
-        match check_database().await {
-            Ok(()) => println!("connected"),
-            Err(e) => println!("error ({})", e),
+    let db_backend = std::env::var("DATABASE_BACKEND")
+        .ok()
+        .unwrap_or_else(|| "postgres".to_string());
+    match db_backend.as_str() {
+        "libsql" | "turso" | "sqlite" => {
+            let path = std::env::var("LIBSQL_PATH")
+                .map(std::path::PathBuf::from)
+                .unwrap_or_else(|_| crate::config::default_libsql_path());
+            if path.exists() {
+                let turso = if std::env::var("LIBSQL_URL").is_ok() {
+                    " + Turso sync"
+                } else {
+                    ""
+                };
+                println!("libSQL ({}{})", path.display(), turso);
+            } else {
+                println!("libSQL (file missing: {})", path.display());
+            }
+        }
+        _ => {
+            if std::env::var("DATABASE_URL").is_ok() {
+                match check_database().await {
+                    Ok(()) => println!("connected (PostgreSQL)"),
+                    Err(e) => println!("error ({})", e),
+                }
+            } else {
+                println!("not configured");
+            }
         }
-    } else {
-        println!("not configured");
     }
 
     // Session / Auth
@@ -43,15 +63,17 @@ pub async fn run_status_command() -> anyhow::Result<()> {
         println!("not found (run `ironclaw onboard`)");
     }
 
-    // Secrets
+    // Secrets (auto-detect from env only; skip keychain probe to avoid
+    // triggering macOS system password dialogs on a simple status check)
     print!("  Secrets:     ");
-    let secrets_configured = settings.secrets_master_key_source != crate::settings::KeySource::None
-        || std::env::var("SECRETS_MASTER_KEY").is_ok()
-        || crate::secrets::keychain::has_master_key().await;
-    if secrets_configured {
-        println!("configured ({:?})", settings.secrets_master_key_source);
+    if std::env::var("SECRETS_MASTER_KEY").is_ok() {
+        println!("configured (env)");
     } else {
-        println!("not configured");
+        // We don't probe the keychain here because get_generic_password()
+        // triggers macOS unlock+authorization dialogs, which is bad UX for
+        // a read-only status command. If onboarding completed with keychain
+        // storage, the key is there; we just can't cheaply verify it.
+        println!("env not set (keychain may be configured)");
     }
 
     // Embeddings
@@ -129,19 +151,18 @@ pub async fn run_status_command() -> anyhow::Result<()> {
         Err(_) => println!("none configured"),
     }
 
-    // Settings path
-    println!("\n  Settings:    {}", Settings::default_path().display());
+    // Config path
+    println!(
+        "\n  Config:      {}",
+        crate::bootstrap::ironclaw_env_path().display()
+    );
 
     Ok(())
 }
 
+#[cfg(feature = "postgres")]
 async fn check_database() -> anyhow::Result<()> {
-    let _ = dotenvy::dotenv();
-    let settings = Settings::load();
-    let url = std::env::var("DATABASE_URL")
-        .ok()
-        .or(settings.database_url)
-        .ok_or_else(|| anyhow::anyhow!("no URL"))?;
+    let url = std::env::var("DATABASE_URL").map_err(|_| anyhow::anyhow!("DATABASE_URL not set"))?;
 
     let config: deadpool_postgres::Config = deadpool_postgres::Config {
         url: Some(url),
@@ -167,6 +188,12 @@ async fn check_database() -> anyhow::Result<()> {
     Ok(())
 }
 
+#[cfg(not(feature = "postgres"))]
+async fn check_database() -> anyhow::Result<()> {
+    // For non-postgres backends, just report configured
+    Ok(())
+}
+
 fn count_wasm_files(dir: &std::path::Path) -> usize {
     std::fs::read_dir(dir)
         .map(|entries| {

src/cli/tool.rs

@@ -11,8 +11,11 @@ use clap::Subcommand;
 use tokio::fs;
 
 use crate::config::Config;
-use crate::history::Store;
-use crate::secrets::{CreateSecretParams, PostgresSecretsStore, SecretsCrypto, SecretsStore};
+#[allow(unused_imports)]
+use crate::db::Database;
+#[cfg(feature = "postgres")]
+use crate::secrets::PostgresSecretsStore;
+use crate::secrets::{CreateSecretParams, SecretsCrypto, SecretsStore};
 use crate::tools::wasm::{CapabilitiesFile, compute_binary_hash};
 
 /// Default tools directory.
@@ -420,11 +423,11 @@ async fn extract_crate_name(cargo_toml: &Path) -> anyhow::Result<String> {
     // Simple TOML parsing for [package] name
     for line in content.lines() {
         let line = line.trim();
-        if line.starts_with("name") {
-            if let Some((_, value)) = line.split_once('=') {
-                let name = value.trim().trim_matches('"').trim_matches('\'');
-                return Ok(name.to_string());
-            }
+        if line.starts_with("name")
+            && let Some((_, value)) = line.split_once('=')
+        {
+            let name = value.trim().trim_matches('"').trim_matches('\'');
+            return Ok(name.to_string());
         }
     }
 
@@ -488,10 +491,10 @@ async fn list_tools(dir: Option<PathBuf>, verbose: bool) -> anyhow::Result<()> {
 
             if has_caps {
                 let caps_path = path.with_extension("capabilities.json");
-                if let Ok(content) = fs::read_to_string(&caps_path).await {
-                    if let Ok(caps) = CapabilitiesFile::from_json(&content) {
-                        print_capabilities_summary(&caps);
-                    }
+                if let Ok(content) = fs::read_to_string(&caps_path).await
+                    && let Ok(caps) = CapabilitiesFile::from_json(&content)
+                {
+                    print_capabilities_summary(&caps);
                 }
             }
             println!();
@@ -604,16 +607,16 @@ fn print_capabilities_summary(caps: &CapabilitiesFile) {
         }
     }
 
-    if let Some(ref secrets) = caps.secrets {
-        if !secrets.allowed_names.is_empty() {
-            parts.push(format!("secrets: {}", secrets.allowed_names.len()));
-        }
+    if let Some(ref secrets) = caps.secrets
+        && !secrets.allowed_names.is_empty()
+    {
+        parts.push(format!("secrets: {}", secrets.allowed_names.len()));
     }
 
-    if let Some(ref ws) = caps.workspace {
-        if !ws.allowed_prefixes.is_empty() {
-            parts.push("workspace: read".to_string());
-        }
+    if let Some(ref ws) = caps.workspace
+        && !ws.allowed_prefixes.is_empty()
+    {
+        parts.push("workspace: read".to_string());
     }
 
     if !parts.is_empty() {
@@ -650,30 +653,30 @@ fn print_capabilities_detail(caps: &CapabilitiesFile) {
         }
     }
 
-    if let Some(ref secrets) = caps.secrets {
-        if !secrets.allowed_names.is_empty() {
-            println!("  Secrets (existence check only):");
-            for name in &secrets.allowed_names {
-                println!("    {}", name);
-            }
+    if let Some(ref secrets) = caps.secrets
+        && !secrets.allowed_names.is_empty()
+    {
+        println!("  Secrets (existence check only):");
+        for name in &secrets.allowed_names {
+            println!("    {}", name);
         }
     }
 
-    if let Some(ref tool_invoke) = caps.tool_invoke {
-        if !tool_invoke.aliases.is_empty() {
-            println!("  Tool aliases:");
-            for (alias, real_name) in &tool_invoke.aliases {
-                println!("    {} -> {}", alias, real_name);
-            }
+    if let Some(ref tool_invoke) = caps.tool_invoke
+        && !tool_invoke.aliases.is_empty()
+    {
+        println!("  Tool aliases:");
+        for (alias, real_name) in &tool_invoke.aliases {
+            println!("    {} -> {}", alias, real_name);
         }
     }
 
-    if let Some(ref ws) = caps.workspace {
-        if !ws.allowed_prefixes.is_empty() {
-            println!("  Workspace read prefixes:");
-            for prefix in &ws.allowed_prefixes {
-                println!("    {}", prefix);
-            }
+    if let Some(ref ws) = caps.workspace
+        && !ws.allowed_prefixes.is_empty()
+    {
+        println!("  Workspace read prefixes:");
+        for prefix in &ws.allowed_prefixes {
+            println!("    {}", prefix);
         }
     }
 }
@@ -722,11 +725,58 @@ async fn auth_tool(name: String, dir: Option<PathBuf>, user_id: String) -> anyho
         )
     })?;
 
-    let store = Store::new(&config.database).await?;
-    store.run_migrations().await?;
-
     let crypto = SecretsCrypto::new(master_key.clone())?;
-    let secrets_store = Arc::new(PostgresSecretsStore::new(store.pool(), Arc::new(crypto)));
+
+    let secrets_store: Arc<dyn SecretsStore + Send + Sync> = {
+        #[cfg(feature = "postgres")]
+        {
+            let store = crate::history::Store::new(&config.database).await?;
+            store.run_migrations().await?;
+            Arc::new(PostgresSecretsStore::new(store.pool(), Arc::new(crypto)))
+        }
+        #[cfg(all(feature = "libsql", not(feature = "postgres")))]
+        {
+            use crate::db::Database as _;
+            use crate::db::libsql_backend::LibSqlBackend;
+            use secrecy::ExposeSecret as _;
+
+            let default_path = crate::config::default_libsql_path();
+            let db_path = config
+                .database
+                .libsql_path
+                .as_deref()
+                .unwrap_or(&default_path);
+
+            let backend = if let Some(ref url) = config.database.libsql_url {
+                let token = config.database.libsql_auth_token.as_ref().ok_or_else(|| {
+                    anyhow::anyhow!("LIBSQL_AUTH_TOKEN is required when LIBSQL_URL is set")
+                })?;
+                LibSqlBackend::new_remote_replica(db_path, url, token.expose_secret())
+                    .await
+                    .map_err(|e| anyhow::anyhow!("{}", e))?
+            } else {
+                LibSqlBackend::new_local(db_path)
+                    .await
+                    .map_err(|e| anyhow::anyhow!("{}", e))?
+            };
+            backend
+                .run_migrations()
+                .await
+                .map_err(|e| anyhow::anyhow!("{}", e))?;
+
+            Arc::new(crate::secrets::LibSqlSecretsStore::new(
+                backend.shared_db(),
+                Arc::new(crypto),
+            ))
+        }
+        #[cfg(not(any(feature = "postgres", feature = "libsql")))]
+        {
+            let _ = crypto;
+            anyhow::bail!(
+                "No database backend available for secrets. Enable 'postgres' or 'libsql' feature."
+            );
+        }
+    };
 
     // Check if already configured
     let already_configured = secrets_store
@@ -752,51 +802,103 @@ async fn auth_tool(name: String, dir: Option<PathBuf>, user_id: String) -> anyho
     }
 
     // Check for environment variable
-    if let Some(ref env_var) = auth.env_var {
-        if let Ok(token) = std::env::var(env_var) {
-            if !token.is_empty() {
-                println!("  Found {} in environment.", env_var);
-                println!();
+    if let Some(ref env_var) = auth.env_var
+        && let Ok(token) = std::env::var(env_var)
+        && !token.is_empty()
+    {
+        println!("  Found {} in environment.", env_var);
+        println!();
 
-                // Validate if endpoint is provided
-                if let Some(ref validation) = auth.validation_endpoint {
-                    print!("  Validating token...");
-                    std::io::stdout().flush()?;
+        // Validate if endpoint is provided
+        if let Some(ref validation) = auth.validation_endpoint {
+            print!("  Validating token...");
+            std::io::stdout().flush()?;
 
-                    match validate_token(&token, validation, &auth.secret_name).await {
-                        Ok(()) => {
-                            println!(" ✓");
-                        }
-                        Err(e) => {
-                            println!(" ✗");
-                            println!("  Validation failed: {}", e);
-                            println!();
-                            println!("  Falling back to manual entry...");
-                            return auth_tool_manual(&secrets_store, &user_id, &auth).await;
-                        }
-                    }
+            match validate_token(&token, validation, &auth.secret_name).await {
+                Ok(()) => {
+                    println!(" ✓");
+                }
+                Err(e) => {
+                    println!(" ✗");
+                    println!("  Validation failed: {}", e);
+                    println!();
+                    println!("  Falling back to manual entry...");
+                    return auth_tool_manual(secrets_store.as_ref(), &user_id, &auth).await;
                 }
-
-                // Save the token
-                save_token(&secrets_store, &user_id, &auth, &token).await?;
-                print_success(display_name);
-                return Ok(());
             }
         }
+
+        // Save the token
+        save_token(secrets_store.as_ref(), &user_id, &auth, &token, None, None).await?;
+        print_success(display_name);
+        return Ok(());
     }
 
     // Check for OAuth configuration
     if let Some(ref oauth) = auth.oauth {
-        return auth_tool_oauth(&secrets_store, &user_id, &auth, oauth).await;
+        // For providers with shared tokens (e.g., all Google tools share google_oauth_token),
+        // combine scopes from all installed tools so one auth covers everything.
+        let combined = combine_provider_scopes(&tools_dir, &auth.secret_name, oauth).await;
+        if combined.scopes.len() > oauth.scopes.len() {
+            let extra = combined.scopes.len() - oauth.scopes.len();
+            println!(
+                "  Including scopes from {} other installed tool(s) sharing this credential.",
+                extra
+            );
+            println!();
+        }
+        return auth_tool_oauth(secrets_store.as_ref(), &user_id, &auth, &combined).await;
     }
 
     // Fall back to manual entry
-    auth_tool_manual(&secrets_store, &user_id, &auth).await
+    auth_tool_manual(secrets_store.as_ref(), &user_id, &auth).await
+}
+
+/// Scan the tools directory for all capabilities files sharing the same secret_name
+/// and combine their OAuth scopes. This way, authing any Google tool requests scopes
+/// for ALL installed Google tools, so one login covers everything.
+async fn combine_provider_scopes(
+    tools_dir: &Path,
+    secret_name: &str,
+    base_oauth: &crate::tools::wasm::OAuthConfigSchema,
+) -> crate::tools::wasm::OAuthConfigSchema {
+    let mut all_scopes: std::collections::HashSet<String> =
+        base_oauth.scopes.iter().cloned().collect();
+
+    if let Ok(mut entries) = tokio::fs::read_dir(tools_dir).await {
+        while let Ok(Some(entry)) = entries.next_entry().await {
+            let path = entry.path();
+            if path.extension().and_then(|e| e.to_str()) != Some("json") {
+                continue;
+            }
+            let name = path
+                .file_name()
+                .and_then(|n| n.to_str())
+                .unwrap_or_default();
+            if !name.ends_with(".capabilities.json") {
+                continue;
+            }
+
+            if let Ok(content) = tokio::fs::read_to_string(&path).await
+                && let Ok(caps) = CapabilitiesFile::from_json(&content)
+                && let Some(auth) = &caps.auth
+                && auth.secret_name == secret_name
+                && let Some(oauth) = &auth.oauth
+            {
+                all_scopes.extend(oauth.scopes.iter().cloned());
+            }
+        }
+    }
+
+    let mut combined = base_oauth.clone();
+    combined.scopes = all_scopes.into_iter().collect();
+    combined.scopes.sort(); // deterministic ordering
+    combined
 }
 
 /// OAuth browser-based login flow.
 async fn auth_tool_oauth(
-    store: &PostgresSecretsStore,
+    store: &(dyn SecretsStore + Send + Sync),
     user_id: &str,
     auth: &crate::tools::wasm::AuthCapabilitySchema,
     oauth: &crate::tools::wasm::OAuthConfigSchema,
@@ -804,12 +906,14 @@ async fn auth_tool_oauth(
     use base64::{Engine, engine::general_purpose::URL_SAFE_NO_PAD};
     use rand::RngCore;
     use sha2::{Digest, Sha256};
-    use tokio::io::{AsyncBufReadExt, AsyncWriteExt, BufReader};
-    use tokio::net::TcpListener;
+
+    use crate::cli::oauth_defaults::{self, OAUTH_CALLBACK_PORT};
 
     let display_name = auth.display_name.as_deref().unwrap_or(&auth.secret_name);
 
-    // Get client_id from config or env
+    // Get client_id: capabilities file > runtime env var > built-in defaults
+    let builtin = oauth_defaults::builtin_credentials(&auth.secret_name);
+
     let client_id = oauth
         .client_id
         .clone()
@@ -819,41 +923,32 @@ async fn auth_tool_oauth(
                 .as_ref()
                 .and_then(|env| std::env::var(env).ok())
         })
+        .or_else(|| builtin.as_ref().map(|c| c.client_id.to_string()))
         .ok_or_else(|| {
             anyhow::anyhow!(
                 "OAuth client_id not configured.\n\
-                 Set it in the capabilities file or via environment variable."
+                 Set {} env var, or build with IRONCLAW_GOOGLE_CLIENT_ID.",
+                oauth.client_id_env.as_deref().unwrap_or("the client_id")
             )
         })?;
 
-    // Get client_secret if provided
-    let client_secret = oauth.client_secret.clone().or_else(|| {
-        oauth
-            .client_secret_env
-            .as_ref()
-            .and_then(|env| std::env::var(env).ok())
-    });
+    // Get client_secret: capabilities file > runtime env var > built-in defaults
+    let client_secret = oauth
+        .client_secret
+        .clone()
+        .or_else(|| {
+            oauth
+                .client_secret_env
+                .as_ref()
+                .and_then(|env| std::env::var(env).ok())
+        })
+        .or_else(|| builtin.as_ref().map(|c| c.client_secret.to_string()));
 
     println!("  Starting OAuth authentication...");
     println!();
 
-    // Find an available port for the callback
-    let mut listener = None;
-    let mut port = 0;
-
-    for p in 9876..=9886 {
-        match TcpListener::bind(format!("127.0.0.1:{}", p)).await {
-            Ok(l) => {
-                listener = Some(l);
-                port = p;
-                break;
-            }
-            Err(_) => continue,
-        }
-    }
-
-    let listener = listener.ok_or_else(|| anyhow::anyhow!("Could not find available port"))?;
-    let redirect_uri = format!("http://localhost:{}/callback", port);
+    let listener = oauth_defaults::bind_callback_listener().await?;
+    let redirect_uri = format!("http://localhost:{}/callback", OAUTH_CALLBACK_PORT);
 
     // Generate PKCE verifier and challenge
     let (code_verifier, code_challenge) = if oauth.use_pkce {
@@ -912,65 +1007,8 @@ async fn auth_tool_oauth(
 
     println!("  Waiting for authorization...");
 
-    // Wait for callback with timeout
-    let timeout = std::time::Duration::from_secs(300);
-    let code = tokio::time::timeout(timeout, async {
-        loop {
-            let (mut socket, _) = listener.accept().await?;
-
-            let mut reader = BufReader::new(&mut socket);
-            let mut request_line = String::new();
-            reader.read_line(&mut request_line).await?;
-
-            // Parse GET /callback?code=xxx HTTP/1.1
-            if let Some(path) = request_line.split_whitespace().nth(1) {
-                if path.starts_with("/callback") {
-                    if let Some(query) = path.split('?').nth(1) {
-                        for param in query.split('&') {
-                            let parts: Vec<&str> = param.splitn(2, '=').collect();
-                            if parts.len() == 2 && parts[0] == "code" {
-                                let code = urlencoding::decode(parts[1])
-                                    .unwrap_or_else(|_| parts[1].into())
-                                    .into_owned();
-
-                                // Send success response
-                                let response = format!(
-                                    "HTTP/1.1 200 OK\r\n\
-                                     Content-Type: text/html\r\n\
-                                     \r\n\
-                                     <!DOCTYPE html><html><body style=\"font-family: sans-serif; \
-                                     display: flex; justify-content: center; align-items: center; \
-                                     height: 100vh; margin: 0; background: #191919; color: white;\">\
-                                     <div style=\"text-align: center;\">\
-                                     <h1>✓ {} Connected!</h1>\
-                                     <p>You can close this window.</p>\
-                                     </div></body></html>",
-                                    display_name
-                                );
-                                let _ = socket.write_all(response.as_bytes()).await;
-                                let _ = socket.shutdown().await;
-
-                                return Ok::<_, anyhow::Error>(code);
-                            }
-                        }
-
-                        // Check for error
-                        if query.contains("error=") {
-                            let response =
-                                "HTTP/1.1 400 Bad Request\r\n\r\nAuthorization denied";
-                            let _ = socket.write_all(response.as_bytes()).await;
-                            return Err(anyhow::anyhow!("Authorization denied by user"));
-                        }
-                    }
-                }
-            }
-
-            let response = "HTTP/1.1 404 Not Found\r\n\r\n";
-            let _ = socket.write_all(response.as_bytes()).await;
-        }
-    })
-    .await
-    .map_err(|_| anyhow::anyhow!("Timed out waiting for authorization"))??;
+    let code =
+        oauth_defaults::wait_for_callback(listener, "/callback", "code", display_name).await?;
 
     println!();
     println!("  Exchanging code for token...");
@@ -1021,8 +1059,19 @@ async fn auth_tool_oauth(
             )
         })?;
 
-    // Save the token
-    save_token(store, user_id, auth, access_token).await?;
+    let refresh_token = token_data.get("refresh_token").and_then(|v| v.as_str());
+    let expires_in = token_data.get("expires_in").and_then(|v| v.as_u64());
+
+    // Save the token (with refresh token and expiry if provided)
+    save_token(
+        store,
+        user_id,
+        auth,
+        access_token,
+        refresh_token,
+        expires_in,
+    )
+    .await?;
 
     // Extract any additional info for display
     let workspace_name = token_data
@@ -1044,7 +1093,7 @@ async fn auth_tool_oauth(
 
 /// Manual token entry flow.
 async fn auth_tool_manual(
-    store: &PostgresSecretsStore,
+    store: &(dyn SecretsStore + Send + Sync),
     user_id: &str,
     auth: &crate::tools::wasm::AuthCapabilitySchema,
 ) -> anyhow::Result<()> {
@@ -1124,8 +1173,8 @@ async fn auth_tool_manual(
         }
     }
 
-    // Save the token
-    save_token(store, user_id, auth, &token).await?;
+    // Save the token (manual path: no refresh token or expiry)
+    save_token(store, user_id, auth, &token, None, None).await?;
     print_success(display_name);
     Ok(())
 }
@@ -1216,11 +1265,16 @@ async fn validate_token(
 }
 
 /// Save token to secrets store.
+///
+/// Optionally stores a refresh token (as `{secret_name}_refresh_token`) and
+/// sets `expires_at` on the access token so the runtime can auto-refresh.
 async fn save_token(
-    store: &PostgresSecretsStore,
+    store: &(dyn SecretsStore + Send + Sync),
     user_id: &str,
     auth: &crate::tools::wasm::AuthCapabilitySchema,
     token: &str,
+    refresh_token: Option<&str>,
+    expires_in: Option<u64>,
 ) -> anyhow::Result<()> {
     let mut params = CreateSecretParams::new(&auth.secret_name, token);
 
@@ -1228,11 +1282,29 @@ async fn save_token(
         params = params.with_provider(provider);
     }
 
+    if let Some(secs) = expires_in {
+        let expires_at = chrono::Utc::now() + chrono::Duration::seconds(secs as i64);
+        params = params.with_expiry(expires_at);
+    }
+
     store
         .create(user_id, params)
         .await
         .map_err(|e| anyhow::anyhow!("Failed to save token: {}", e))?;
 
+    // Store refresh token separately (no expiry, it's long-lived)
+    if let Some(rt) = refresh_token {
+        let refresh_name = format!("{}_refresh_token", auth.secret_name);
+        let mut refresh_params = CreateSecretParams::new(&refresh_name, rt);
+        if let Some(ref provider) = auth.provider {
+            refresh_params = refresh_params.with_provider(provider);
+        }
+        store
+            .create(user_id, refresh_params)
+            .await
+            .map_err(|e| anyhow::anyhow!("Failed to save refresh token: {}", e))?;
+    }
+
     Ok(())
 }
 

src/config.rs

@@ -1,11 +1,13 @@
 //! Configuration for IronClaw.
 //!
 //! Settings are loaded with priority: env var > database > default.
-//! The database replaces the old `settings.json` file for all settings
-//! except the 4 bootstrap fields (database_url, pool_size, secrets key
-//! source, onboard_completed) which live in `~/.ironclaw/bootstrap.json`.
+//! `DATABASE_URL` lives in `~/.ironclaw/.env` (loaded via dotenvy early
+//! in startup). Everything else comes from env vars, the DB settings
+//! table, or auto-detection.
 
+use std::collections::HashMap;
 use std::path::PathBuf;
+use std::sync::OnceLock;
 use std::time::Duration;
 
 use secrecy::{ExposeSecret, SecretString};
@@ -13,6 +15,13 @@ use secrecy::{ExposeSecret, SecretString};
 use crate::error::ConfigError;
 use crate::settings::Settings;
 
+/// Thread-safe overlay for injected env vars (secrets loaded from DB).
+///
+/// Used by `inject_llm_keys_from_secrets()` to make API keys available to
+/// `optional_env()` without unsafe `set_var` calls. `optional_env()` checks
+/// real env vars first, then falls back to this overlay.
+static INJECTED_VARS: OnceLock<HashMap<String, String>> = OnceLock::new();
+
 /// Main configuration for the agent.
 #[derive(Debug, Clone)]
 pub struct Config {
@@ -38,11 +47,11 @@ impl Config {
     /// Priority: env var > DB settings > default.
     /// This is the primary way to load config after DB is connected.
     pub async fn from_db(
-        store: &crate::history::Store,
+        store: &dyn crate::db::Database,
         user_id: &str,
-        bootstrap: &crate::bootstrap::BootstrapConfig,
     ) -> Result<Self, ConfigError> {
         let _ = dotenvy::dotenv();
+        crate::bootstrap::load_ironclaw_env();
 
         // Load all settings from DB into a Settings struct
         let db_settings = match store.get_all_settings(user_id).await {
@@ -53,7 +62,7 @@ impl Config {
             }
         };
 
-        Self::build(bootstrap, &db_settings).await
+        Self::build(&db_settings).await
     }
 
     /// Load configuration from environment variables only (no database).
@@ -61,20 +70,20 @@ impl Config {
     /// Used during early startup before the database is connected,
     /// and by CLI commands that don't have DB access.
     /// Falls back to legacy `settings.json` on disk if present.
+    ///
+    /// Loads both `./.env` (standard, higher priority) and `~/.ironclaw/.env`
+    /// (lower priority) via dotenvy, which never overwrites existing vars.
     pub async fn from_env() -> Result<Self, ConfigError> {
         let _ = dotenvy::dotenv();
-        let bootstrap = crate::bootstrap::BootstrapConfig::load();
+        crate::bootstrap::load_ironclaw_env();
         let settings = Settings::load();
-        Self::build(&bootstrap, &settings).await
+        Self::build(&settings).await
     }
 
-    /// Build config from bootstrap + settings (shared by from_env and from_db).
-    async fn build(
-        bootstrap: &crate::bootstrap::BootstrapConfig,
-        settings: &Settings,
-    ) -> Result<Self, ConfigError> {
+    /// Build config from settings (shared by from_env and from_db).
+    async fn build(settings: &Settings) -> Result<Self, ConfigError> {
         Ok(Self {
-            database: DatabaseConfig::resolve(bootstrap)?,
+            database: DatabaseConfig::resolve()?,
             llm: LlmConfig::resolve(settings)?,
             embeddings: EmbeddingsConfig::resolve(settings)?,
             tunnel: TunnelConfig::resolve(settings)?,
@@ -82,7 +91,7 @@ impl Config {
             agent: AgentConfig::resolve(settings)?,
             safety: SafetyConfig::resolve()?,
             wasm: WasmConfig::resolve()?,
-            secrets: SecretsConfig::resolve(bootstrap).await?,
+            secrets: SecretsConfig::resolve().await?,
             builder: BuilderModeConfig::resolve()?,
             heartbeat: HeartbeatConfig::resolve(settings)?,
             routines: RoutineConfig::resolve()?,
@@ -107,13 +116,13 @@ impl TunnelConfig {
         let public_url = optional_env("TUNNEL_URL")?
             .or_else(|| settings.tunnel.public_url.clone().filter(|s| !s.is_empty()));
 
-        if let Some(ref url) = public_url {
-            if !url.starts_with("https://") {
-                return Err(ConfigError::InvalidValue {
-                    key: "TUNNEL_URL".to_string(),
-                    message: "must start with https:// (webhooks require HTTPS)".to_string(),
-                });
-            }
+        if let Some(ref url) = public_url
+            && !url.starts_with("https://")
+        {
+            return Err(ConfigError::InvalidValue {
+                key: "TUNNEL_URL".to_string(),
+                message: "must start with https:// (webhooks require HTTPS)".to_string(),
+            });
         }
 
         Ok(Self { public_url })
@@ -134,35 +143,113 @@ impl TunnelConfig {
     }
 }
 
+/// Which database backend to use.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
+pub enum DatabaseBackend {
+    /// PostgreSQL via deadpool-postgres (default).
+    #[default]
+    Postgres,
+    /// libSQL/Turso embedded database.
+    LibSql,
+}
+
+impl std::fmt::Display for DatabaseBackend {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Self::Postgres => write!(f, "postgres"),
+            Self::LibSql => write!(f, "libsql"),
+        }
+    }
+}
+
+impl std::str::FromStr for DatabaseBackend {
+    type Err = String;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        match s.to_lowercase().as_str() {
+            "postgres" | "postgresql" | "pg" => Ok(Self::Postgres),
+            "libsql" | "turso" | "sqlite" => Ok(Self::LibSql),
+            _ => Err(format!(
+                "invalid database backend '{}', expected 'postgres' or 'libsql'",
+                s
+            )),
+        }
+    }
+}
+
 /// Database configuration.
 #[derive(Debug, Clone)]
 pub struct DatabaseConfig {
+    /// Which backend to use (default: Postgres).
+    pub backend: DatabaseBackend,
+
+    // -- PostgreSQL fields --
     pub url: SecretString,
     pub pool_size: usize,
+
+    // -- libSQL fields --
+    /// Path to local libSQL database file (default: ~/.ironclaw/ironclaw.db).
+    pub libsql_path: Option<PathBuf>,
+    /// Turso cloud URL for remote sync (optional).
+    pub libsql_url: Option<String>,
+    /// Turso auth token (required when libsql_url is set).
+    pub libsql_auth_token: Option<SecretString>,
 }
 
 impl DatabaseConfig {
-    fn resolve(bootstrap: &crate::bootstrap::BootstrapConfig) -> Result<Self, ConfigError> {
+    fn resolve() -> Result<Self, ConfigError> {
+        let backend: DatabaseBackend = if let Some(b) = optional_env("DATABASE_BACKEND")? {
+            b.parse().map_err(|e| ConfigError::InvalidValue {
+                key: "DATABASE_BACKEND".to_string(),
+                message: e,
+            })?
+        } else {
+            DatabaseBackend::default()
+        };
+
+        // PostgreSQL URL is required only when using the postgres backend.
+        // For libsql backend, default to an empty placeholder.
+        // DATABASE_URL is loaded from ~/.ironclaw/.env via dotenvy early in startup.
         let url = optional_env("DATABASE_URL")?
-            .or_else(|| bootstrap.database_url.clone())
+            .or_else(|| {
+                if backend == DatabaseBackend::LibSql {
+                    Some("unused://libsql".to_string())
+                } else {
+                    None
+                }
+            })
             .ok_or_else(|| ConfigError::MissingRequired {
                 key: "database_url".to_string(),
                 hint: "Run 'ironclaw onboard' or set DATABASE_URL environment variable".to_string(),
             })?;
 
-        let pool_size = optional_env("DATABASE_POOL_SIZE")?
-            .map(|s| s.parse())
-            .transpose()
-            .map_err(|e| ConfigError::InvalidValue {
-                key: "DATABASE_POOL_SIZE".to_string(),
-                message: format!("must be a positive integer: {e}"),
-            })?
-            .or(bootstrap.database_pool_size)
-            .unwrap_or(10);
+        let pool_size = parse_optional_env("DATABASE_POOL_SIZE", 10)?;
+
+        let libsql_path = optional_env("LIBSQL_PATH")?.map(PathBuf::from).or_else(|| {
+            if backend == DatabaseBackend::LibSql {
+                Some(default_libsql_path())
+            } else {
+                None
+            }
+        });
+
+        let libsql_url = optional_env("LIBSQL_URL")?;
+        let libsql_auth_token = optional_env("LIBSQL_AUTH_TOKEN")?.map(SecretString::from);
+
+        if libsql_url.is_some() && libsql_auth_token.is_none() {
+            return Err(ConfigError::MissingRequired {
+                key: "LIBSQL_AUTH_TOKEN".to_string(),
+                hint: "LIBSQL_AUTH_TOKEN is required when LIBSQL_URL is set".to_string(),
+            });
+        }
 
         Ok(Self {
+            backend,
             url: SecretString::from(url),
             pool_size,
+            libsql_path,
+            libsql_url,
+            libsql_auth_token,
         })
     }
 
@@ -172,6 +259,14 @@ impl DatabaseConfig {
     }
 }
 
+/// Default libSQL database path (~/.ironclaw/ironclaw.db).
+pub fn default_libsql_path() -> PathBuf {
+    dirs::home_dir()
+        .unwrap_or_else(|| PathBuf::from("."))
+        .join(".ironclaw")
+        .join("ironclaw.db")
+}
+
 /// Which LLM backend to use.
 ///
 /// Defaults to `NearAi` to keep IronClaw close to the NEAR ecosystem.
@@ -302,6 +397,9 @@ impl std::str::FromStr for NearAiApiMode {
 pub struct NearAiConfig {
     /// Model to use (e.g., "claude-3-5-sonnet-20241022", "gpt-4o")
     pub model: String,
+    /// Cheap/fast model for lightweight tasks (heartbeat, routing, evaluation).
+    /// Falls back to the main model if not set.
+    pub cheap_model: Option<String>,
     /// Base URL for the NEAR AI API (default: https://api.near.ai)
     pub base_url: String,
     /// Base URL for auth/refresh endpoints (default: https://private.near.ai)
@@ -312,16 +410,44 @@ pub struct NearAiConfig {
     pub api_mode: NearAiApiMode,
     /// API key for cloud-api (required for chat_completions mode)
     pub api_key: Option<SecretString>,
+    /// Optional fallback model for failover (default: None).
+    /// When set, a secondary provider is created with this model and wrapped
+    /// in a `FailoverProvider` so transient errors on the primary model
+    /// automatically fall through to the fallback.
+    pub fallback_model: Option<String>,
+    /// Maximum number of retries for transient errors (default: 3).
+    /// With the default of 3, the provider makes up to 4 total attempts
+    /// (1 initial + 3 retries) before giving up.
+    pub max_retries: u32,
+    /// Cooldown duration in seconds for the failover provider (default: 300).
+    /// When a provider accumulates enough consecutive failures it is skipped
+    /// for this many seconds.
+    pub failover_cooldown_secs: u64,
+    /// Number of consecutive retryable failures before a provider enters
+    /// cooldown (default: 3).
+    pub failover_cooldown_threshold: u32,
 }
 
 impl LlmConfig {
     fn resolve(settings: &Settings) -> Result<Self, ConfigError> {
-        // Determine backend (default: NearAi)
+        // Determine backend: env var > settings > default (NearAi)
         let backend: LlmBackend = if let Some(b) = optional_env("LLM_BACKEND")? {
             b.parse().map_err(|e| ConfigError::InvalidValue {
                 key: "LLM_BACKEND".to_string(),
                 message: e,
             })?
+        } else if let Some(ref b) = settings.llm_backend {
+            match b.parse() {
+                Ok(backend) => backend,
+                Err(e) => {
+                    tracing::warn!(
+                        "Invalid llm_backend '{}' in settings: {}. Using default NearAi.",
+                        b,
+                        e
+                    );
+                    LlmBackend::NearAi
+                }
+            }
         } else {
             LlmBackend::NearAi
         };
@@ -347,6 +473,7 @@ impl LlmConfig {
                     "fireworks::accounts/fireworks/models/llama4-maverick-instruct-basic"
                         .to_string()
                 }),
+            cheap_model: optional_env("NEARAI_CHEAP_MODEL")?,
             base_url: optional_env("NEARAI_BASE_URL")?
                 .unwrap_or_else(|| "https://cloud-api.near.ai".to_string()),
             auth_base_url: optional_env("NEARAI_AUTH_URL")?
@@ -356,6 +483,10 @@ impl LlmConfig {
                 .unwrap_or_else(default_session_path),
             api_mode,
             api_key: nearai_api_key,
+            fallback_model: optional_env("NEARAI_FALLBACK_MODEL")?,
+            max_retries: parse_optional_env("NEARAI_MAX_RETRIES", 3)?,
+            failover_cooldown_secs: parse_optional_env("LLM_FAILOVER_COOLDOWN_SECS", 300)?,
+            failover_cooldown_threshold: parse_optional_env("LLM_FAILOVER_THRESHOLD", 3)?,
         };
 
         // Resolve provider-specific configs based on backend
@@ -388,6 +519,7 @@ impl LlmConfig {
 
         let ollama = if backend == LlmBackend::Ollama {
             let base_url = optional_env("OLLAMA_BASE_URL")?
+                .or_else(|| settings.ollama_base_url.clone())
                 .unwrap_or_else(|| "http://localhost:11434".to_string());
             let model = optional_env("OLLAMA_MODEL")?.unwrap_or_else(|| "llama3".to_string());
             Some(OllamaConfig { base_url, model })
@@ -396,8 +528,9 @@ impl LlmConfig {
         };
 
         let openai_compatible = if backend == LlmBackend::OpenAiCompatible {
-            let base_url =
-                optional_env("LLM_BASE_URL")?.ok_or_else(|| ConfigError::MissingRequired {
+            let base_url = optional_env("LLM_BASE_URL")?
+                .or_else(|| settings.openai_compatible_base_url.clone())
+                .ok_or_else(|| ConfigError::MissingRequired {
                     key: "LLM_BASE_URL".to_string(),
                     hint: "Set LLM_BASE_URL when LLM_BACKEND=openai_compatible".to_string(),
                 })?;
@@ -767,52 +900,41 @@ impl std::fmt::Debug for SecretsConfig {
     }
 }
 
+/// Process-wide cache for the keychain master key.
+///
+/// Avoids re-prompting the OS keychain on every `SecretsConfig::resolve()` call
+/// (e.g. `Config::from_env()` then `Config::from_db()`). Thread-safe alternative
+/// to caching in a process env var.
 impl SecretsConfig {
-    async fn resolve(bootstrap: &crate::bootstrap::BootstrapConfig) -> Result<Self, ConfigError> {
+    /// Auto-detect secrets master key from env var, then OS keychain.
+    ///
+    /// Sequential probe: SECRETS_MASTER_KEY env var first, then OS keychain.
+    /// No saved "source" needed; just try each source in order.
+    async fn resolve() -> Result<Self, ConfigError> {
         use crate::settings::KeySource;
 
         let (master_key, source) = if let Some(env_key) = optional_env("SECRETS_MASTER_KEY")? {
             (Some(SecretString::from(env_key)), KeySource::Env)
         } else {
-            match bootstrap.secrets_master_key_source {
-                KeySource::Keychain => {
-                    // Try to load from OS keychain (async on Linux)
-                    match crate::secrets::keychain::get_master_key().await {
-                        Ok(key_bytes) => {
-                            let key_hex: String =
-                                key_bytes.iter().map(|b| format!("{:02x}", b)).collect();
-                            (Some(SecretString::from(key_hex)), KeySource::Keychain)
-                        }
-                        Err(_) => {
-                            // Keychain configured but key not found
-                            // This might happen if keychain was cleared
-                            tracing::warn!(
-                                "Secrets configured for keychain but key not found. \
-                                 Run 'ironclaw onboard' to reconfigure."
-                            );
-                            (None, KeySource::None)
-                        }
-                    }
+            // Probe the OS keychain; if a key is stored, use it
+            match crate::secrets::keychain::get_master_key().await {
+                Ok(key_bytes) => {
+                    let key_hex: String = key_bytes.iter().map(|b| format!("{:02x}", b)).collect();
+                    (Some(SecretString::from(key_hex)), KeySource::Keychain)
                 }
-                KeySource::Env => {
-                    tracing::warn!(
-                        "Secrets configured for env var but SECRETS_MASTER_KEY not set."
-                    );
-                    (None, KeySource::None)
-                }
-                KeySource::None => (None, KeySource::None),
+                Err(_) => (None, KeySource::None),
             }
         };
 
         let enabled = master_key.is_some();
 
-        if let Some(ref key) = master_key {
-            if key.expose_secret().len() < 32 {
-                return Err(ConfigError::InvalidValue {
-                    key: "SECRETS_MASTER_KEY".to_string(),
-                    message: "must be at least 32 bytes for AES-256-GCM".to_string(),
-                });
-            }
+        if let Some(ref key) = master_key
+            && key.expose_secret().len() < 32
+        {
+            return Err(ConfigError::InvalidValue {
+                key: "SECRETS_MASTER_KEY".to_string(),
+                message: "must be at least 32 bytes for AES-256-GCM".to_string(),
+            });
         }
 
         Ok(Self {
@@ -1174,6 +1296,36 @@ pub struct ClaudeCodeConfig {
     pub max_turns: u32,
     /// Memory limit in MB for Claude Code containers (heavier than workers).
     pub memory_limit_mb: u64,
+    /// Allowed tool patterns for Claude Code permission settings.
+    ///
+    /// Written to `/workspace/.claude/settings.json` before spawning the CLI.
+    /// Provides defense-in-depth: only explicitly listed tools are auto-approved.
+    /// Any new/unknown tools would require interactive approval (which times out
+    /// in the non-interactive container, failing safely).
+    ///
+    /// Patterns follow Claude Code syntax: `"Bash(*)"`, `"Read"`, `"Edit(*)"`, etc.
+    pub allowed_tools: Vec<String>,
+}
+
+/// Default allowed tools for Claude Code inside containers.
+///
+/// These cover all standard Claude Code tools needed for autonomous operation.
+/// The Docker container provides the primary security boundary; this allowlist
+/// provides defense-in-depth by preventing any future unknown tools from being
+/// silently auto-approved.
+fn default_claude_code_allowed_tools() -> Vec<String> {
+    [
+        "Bash(*)",
+        "Read",
+        "Edit(*)",
+        "Glob",
+        "Grep",
+        "WebFetch(*)",
+        "Task(*)",
+    ]
+    .into_iter()
+    .map(String::from)
+    .collect()
 }
 
 impl Default for ClaudeCodeConfig {
@@ -1186,11 +1338,24 @@ impl Default for ClaudeCodeConfig {
             model: "sonnet".to_string(),
             max_turns: 50,
             memory_limit_mb: 4096,
+            allowed_tools: default_claude_code_allowed_tools(),
         }
     }
 }
 
 impl ClaudeCodeConfig {
+    /// Load from environment variables only (used inside containers where
+    /// there is no database or full config).
+    pub fn from_env() -> Self {
+        match Self::resolve() {
+            Ok(c) => c,
+            Err(e) => {
+                tracing::warn!("Failed to resolve ClaudeCodeConfig: {e}, using defaults");
+                Self::default()
+            }
+        }
+    }
+
     fn resolve() -> Result<Self, ConfigError> {
         let defaults = Self::default();
         Ok(Self {
@@ -1211,21 +1376,76 @@ impl ClaudeCodeConfig {
                 "CLAUDE_CODE_MEMORY_LIMIT_MB",
                 defaults.memory_limit_mb,
             )?,
+            allowed_tools: optional_env("CLAUDE_CODE_ALLOWED_TOOLS")?
+                .map(|s| {
+                    s.split(',')
+                        .map(|t| t.trim().to_string())
+                        .filter(|t| !t.is_empty())
+                        .collect()
+                })
+                .unwrap_or(defaults.allowed_tools),
         })
     }
 }
 
+/// Load API keys from the encrypted secrets store into a thread-safe overlay.
+///
+/// This bridges the gap between secrets stored during onboarding and the
+/// env-var-first resolution in `LlmConfig::resolve()`. Keys in the overlay
+/// are read by `optional_env()` before falling back to `std::env::var()`,
+/// so explicit env vars always win.
+pub async fn inject_llm_keys_from_secrets(
+    secrets: &dyn crate::secrets::SecretsStore,
+    user_id: &str,
+) {
+    let mappings = [
+        ("llm_openai_api_key", "OPENAI_API_KEY"),
+        ("llm_anthropic_api_key", "ANTHROPIC_API_KEY"),
+        ("llm_compatible_api_key", "LLM_API_KEY"),
+    ];
+
+    let mut injected = HashMap::new();
+
+    for (secret_name, env_var) in mappings {
+        match std::env::var(env_var) {
+            Ok(val) if !val.is_empty() => continue,
+            _ => {}
+        }
+        match secrets.get_decrypted(user_id, secret_name).await {
+            Ok(decrypted) => {
+                injected.insert(env_var.to_string(), decrypted.expose().to_string());
+                tracing::debug!("Loaded secret '{}' for env var '{}'", secret_name, env_var);
+            }
+            Err(_) => {
+                // Secret doesn't exist, that's fine
+            }
+        }
+    }
+
+    let _ = INJECTED_VARS.set(injected);
+}
+
 // Helper functions
 
 fn optional_env(key: &str) -> Result<Option<String>, ConfigError> {
+    // Check real env vars first (always win over injected secrets)
     match std::env::var(key) {
-        Ok(val) if val.is_empty() => Ok(None),
-        Ok(val) => Ok(Some(val)),
-        Err(std::env::VarError::NotPresent) => Ok(None),
-        Err(e) => Err(ConfigError::ParseError(format!(
-            "failed to read {key}: {e}"
-        ))),
+        Ok(val) if val.is_empty() => {}
+        Ok(val) => return Ok(Some(val)),
+        Err(std::env::VarError::NotPresent) => {}
+        Err(e) => {
+            return Err(ConfigError::ParseError(format!(
+                "failed to read {key}: {e}"
+            )));
+        }
     }
+
+    // Fall back to thread-safe overlay (secrets injected from DB)
+    if let Some(val) = INJECTED_VARS.get().and_then(|map| map.get(key)) {
+        return Ok(Some(val.clone()));
+    }
+
+    Ok(None)
 }
 
 fn parse_optional_env<T>(key: &str, default: T) -> Result<T, ConfigError>

src/context/manager.rs

@@ -45,20 +45,21 @@ impl ContextManager {
         title: impl Into<String>,
         description: impl Into<String>,
     ) -> Result<Uuid, JobError> {
-        let contexts = self.contexts.read().await;
+        // Hold write lock for the entire check-insert to prevent TOCTOU races
+        // where two concurrent calls both pass the active_count check.
+        let mut contexts = self.contexts.write().await;
         let active_count = contexts.values().filter(|c| c.state.is_active()).count();
 
         if active_count >= self.max_jobs {
             return Err(JobError::MaxJobsExceeded { max: self.max_jobs });
         }
-        drop(contexts);
 
         let context = JobContext::with_user(user_id, title, description);
         let job_id = context.job_id;
+        contexts.insert(job_id, context);
+        drop(contexts);
 
         let memory = Memory::new(job_id);
-
-        self.contexts.write().await.insert(job_id, context);
         self.memories.write().await.insert(job_id, memory);
 
         Ok(job_id)

src/context/state.rs

@@ -119,6 +119,10 @@ pub struct JobContext {
     pub estimated_duration: Option<Duration>,
     /// Actual cost so far.
     pub actual_cost: Decimal,
+    /// Total tokens consumed by LLM calls in this job.
+    pub total_tokens_used: u64,
+    /// Maximum tokens allowed per job (0 = unlimited).
+    pub max_tokens: u64,
     /// When the job was created.
     pub created_at: DateTime<Utc>,
     /// When the job was started.
@@ -159,6 +163,8 @@ impl JobContext {
             estimated_cost: None,
             estimated_duration: None,
             actual_cost: Decimal::ZERO,
+            total_tokens_used: 0,
+            max_tokens: 0,
             created_at: Utc::now(),
             started_at: None,
             completed_at: None,
@@ -189,6 +195,14 @@ impl JobContext {
         };
 
         self.transitions.push(transition);
+
+        // Cap transition history to prevent unbounded memory growth
+        const MAX_TRANSITIONS: usize = 200;
+        if self.transitions.len() > MAX_TRANSITIONS {
+            let drain_count = self.transitions.len() - MAX_TRANSITIONS;
+            self.transitions.drain(..drain_count);
+        }
+
         self.state = new_state;
 
         // Update timestamps
@@ -210,6 +224,29 @@ impl JobContext {
         self.actual_cost += cost;
     }
 
+    /// Record token usage from an LLM call. Returns an error string if the
+    /// token budget has been exceeded after this addition.
+    pub fn add_tokens(&mut self, tokens: u64) -> Result<(), String> {
+        self.total_tokens_used += tokens;
+        if self.max_tokens > 0 && self.total_tokens_used > self.max_tokens {
+            Err(format!(
+                "Token budget exceeded: used {} of {} allowed tokens",
+                self.total_tokens_used, self.max_tokens
+            ))
+        } else {
+            Ok(())
+        }
+    }
+
+    /// Check whether the monetary budget has been exceeded.
+    pub fn budget_exceeded(&self) -> bool {
+        if let Some(ref budget) = self.budget {
+            self.actual_cost > *budget
+        } else {
+            false
+        }
+    }
+
     /// Get the duration since the job started.
     pub fn elapsed(&self) -> Option<Duration> {
         self.started_at.map(|start| {
@@ -274,6 +311,57 @@ mod tests {
         assert_eq!(ctx.state, JobState::Completed);
     }
 
+    #[test]
+    fn test_transition_history_capped() {
+        let mut ctx = JobContext::new("Test", "Transition cap test");
+        // Cycle through Pending -> InProgress -> Stuck -> InProgress -> Stuck ...
+        ctx.transition_to(JobState::InProgress, None).unwrap();
+        for i in 0..250 {
+            ctx.mark_stuck(format!("stuck {}", i)).unwrap();
+            ctx.attempt_recovery().unwrap();
+        }
+        // 1 initial + 250*2 = 501 transitions, should be capped at 200
+        assert!(
+            ctx.transitions.len() <= 200,
+            "transitions should be capped at 200, got {}",
+            ctx.transitions.len()
+        );
+    }
+
+    #[test]
+    fn test_add_tokens_enforces_budget() {
+        let mut ctx = JobContext::new("Test", "Budget test");
+        ctx.max_tokens = 1000;
+        assert!(ctx.add_tokens(500).is_ok());
+        assert_eq!(ctx.total_tokens_used, 500);
+        assert!(ctx.add_tokens(600).is_err());
+        assert_eq!(ctx.total_tokens_used, 1100); // tokens still recorded
+    }
+
+    #[test]
+    fn test_add_tokens_unlimited() {
+        let mut ctx = JobContext::new("Test", "No budget");
+        // max_tokens = 0 means unlimited
+        assert!(ctx.add_tokens(1_000_000).is_ok());
+    }
+
+    #[test]
+    fn test_budget_exceeded() {
+        let mut ctx = JobContext::new("Test", "Money test");
+        ctx.budget = Some(Decimal::new(100, 0)); // $100
+        assert!(!ctx.budget_exceeded());
+        ctx.add_cost(Decimal::new(50, 0));
+        assert!(!ctx.budget_exceeded());
+        ctx.add_cost(Decimal::new(60, 0));
+        assert!(ctx.budget_exceeded());
+    }
+
+    #[test]
+    fn test_budget_exceeded_none() {
+        let ctx = JobContext::new("Test", "No budget");
+        assert!(!ctx.budget_exceeded()); // No budget = never exceeded
+    }
+
     #[test]
     fn test_stuck_recovery() {
         let mut ctx = JobContext::new("Test", "Test job");

src/db/libsql_migrations.rs

@@ -0,0 +1,549 @@
+//! SQLite-dialect migrations for the libSQL/Turso backend.
+//!
+//! Consolidates all PostgreSQL migrations (V1-V8) into a single SQLite-compatible
+//! schema. Run once on database creation; idempotent via `IF NOT EXISTS`.
+
+/// Consolidated schema for libSQL.
+///
+/// Translates PostgreSQL types and features:
+/// - `UUID` -> `TEXT` (store as hex string)
+/// - `TIMESTAMPTZ` -> `TEXT` (ISO-8601)
+/// - `JSONB` -> `TEXT` (JSON encoded)
+/// - `BYTEA` -> `BLOB`
+/// - `NUMERIC` -> `TEXT` (preserve precision for rust_decimal)
+/// - `TEXT[]` -> `TEXT` (JSON array)
+/// - `VECTOR(1536)` -> `F32_BLOB(1536)` (libsql native)
+/// - `TSVECTOR` -> FTS5 virtual table
+/// - `BIGSERIAL` -> `INTEGER PRIMARY KEY AUTOINCREMENT`
+/// - PL/pgSQL functions -> SQLite triggers
+pub const SCHEMA: &str = r#"
+
+-- ==================== Migration tracking ====================
+
+CREATE TABLE IF NOT EXISTS _migrations (
+    version INTEGER PRIMARY KEY,
+    name TEXT NOT NULL,
+    applied_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+-- ==================== Conversations ====================
+
+CREATE TABLE IF NOT EXISTS conversations (
+    id TEXT PRIMARY KEY,
+    channel TEXT NOT NULL,
+    user_id TEXT NOT NULL,
+    thread_id TEXT,
+    started_at TEXT NOT NULL DEFAULT (datetime('now')),
+    last_activity TEXT NOT NULL DEFAULT (datetime('now')),
+    metadata TEXT NOT NULL DEFAULT '{}'
+);
+
+CREATE INDEX IF NOT EXISTS idx_conversations_channel ON conversations(channel);
+CREATE INDEX IF NOT EXISTS idx_conversations_user ON conversations(user_id);
+CREATE INDEX IF NOT EXISTS idx_conversations_last_activity ON conversations(last_activity);
+
+CREATE TABLE IF NOT EXISTS conversation_messages (
+    id TEXT PRIMARY KEY,
+    conversation_id TEXT NOT NULL REFERENCES conversations(id) ON DELETE CASCADE,
+    role TEXT NOT NULL,
+    content TEXT NOT NULL,
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_conversation_messages_conversation
+    ON conversation_messages(conversation_id);
+
+-- ==================== Agent Jobs ====================
+
+CREATE TABLE IF NOT EXISTS agent_jobs (
+    id TEXT PRIMARY KEY,
+    marketplace_job_id TEXT,
+    conversation_id TEXT REFERENCES conversations(id),
+    title TEXT NOT NULL,
+    description TEXT NOT NULL,
+    category TEXT,
+    status TEXT NOT NULL,
+    source TEXT NOT NULL,
+    user_id TEXT NOT NULL DEFAULT 'default',
+    project_dir TEXT,
+    job_mode TEXT NOT NULL DEFAULT 'worker',
+    budget_amount TEXT,
+    budget_token TEXT,
+    bid_amount TEXT,
+    estimated_cost TEXT,
+    estimated_time_secs INTEGER,
+    estimated_value TEXT,
+    actual_cost TEXT,
+    actual_time_secs INTEGER,
+    success INTEGER,
+    failure_reason TEXT,
+    stuck_since TEXT,
+    repair_attempts INTEGER NOT NULL DEFAULT 0,
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    started_at TEXT,
+    completed_at TEXT
+);
+
+CREATE INDEX IF NOT EXISTS idx_agent_jobs_status ON agent_jobs(status);
+CREATE INDEX IF NOT EXISTS idx_agent_jobs_marketplace ON agent_jobs(marketplace_job_id);
+CREATE INDEX IF NOT EXISTS idx_agent_jobs_conversation ON agent_jobs(conversation_id);
+CREATE INDEX IF NOT EXISTS idx_agent_jobs_source ON agent_jobs(source);
+CREATE INDEX IF NOT EXISTS idx_agent_jobs_user ON agent_jobs(user_id);
+CREATE INDEX IF NOT EXISTS idx_agent_jobs_created ON agent_jobs(created_at DESC);
+
+CREATE TABLE IF NOT EXISTS job_actions (
+    id TEXT PRIMARY KEY,
+    job_id TEXT NOT NULL REFERENCES agent_jobs(id) ON DELETE CASCADE,
+    sequence_num INTEGER NOT NULL,
+    tool_name TEXT NOT NULL,
+    input TEXT NOT NULL,
+    output_raw TEXT,
+    output_sanitized TEXT,
+    sanitization_warnings TEXT,
+    cost TEXT,
+    duration_ms INTEGER,
+    success INTEGER NOT NULL,
+    error_message TEXT,
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    UNIQUE(job_id, sequence_num)
+);
+
+CREATE INDEX IF NOT EXISTS idx_job_actions_job_id ON job_actions(job_id);
+CREATE INDEX IF NOT EXISTS idx_job_actions_tool ON job_actions(tool_name);
+
+-- ==================== Dynamic Tools ====================
+
+CREATE TABLE IF NOT EXISTS dynamic_tools (
+    id TEXT PRIMARY KEY,
+    name TEXT NOT NULL UNIQUE,
+    description TEXT NOT NULL,
+    parameters_schema TEXT NOT NULL,
+    code TEXT NOT NULL,
+    sandbox_config TEXT NOT NULL,
+    created_by_job_id TEXT REFERENCES agent_jobs(id),
+    success_count INTEGER NOT NULL DEFAULT 0,
+    failure_count INTEGER NOT NULL DEFAULT 0,
+    last_error TEXT,
+    status TEXT NOT NULL DEFAULT 'active',
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    updated_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_dynamic_tools_status ON dynamic_tools(status);
+CREATE INDEX IF NOT EXISTS idx_dynamic_tools_name ON dynamic_tools(name);
+
+-- ==================== LLM Calls ====================
+
+CREATE TABLE IF NOT EXISTS llm_calls (
+    id TEXT PRIMARY KEY,
+    job_id TEXT REFERENCES agent_jobs(id) ON DELETE CASCADE,
+    conversation_id TEXT REFERENCES conversations(id),
+    provider TEXT NOT NULL,
+    model TEXT NOT NULL,
+    input_tokens INTEGER NOT NULL,
+    output_tokens INTEGER NOT NULL,
+    cost TEXT NOT NULL,
+    purpose TEXT,
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_llm_calls_job ON llm_calls(job_id);
+CREATE INDEX IF NOT EXISTS idx_llm_calls_conversation ON llm_calls(conversation_id);
+CREATE INDEX IF NOT EXISTS idx_llm_calls_provider ON llm_calls(provider);
+
+-- ==================== Estimation ====================
+
+CREATE TABLE IF NOT EXISTS estimation_snapshots (
+    id TEXT PRIMARY KEY,
+    job_id TEXT NOT NULL REFERENCES agent_jobs(id) ON DELETE CASCADE,
+    category TEXT NOT NULL,
+    tool_names TEXT NOT NULL DEFAULT '[]',
+    estimated_cost TEXT NOT NULL,
+    actual_cost TEXT,
+    estimated_time_secs INTEGER NOT NULL,
+    actual_time_secs INTEGER,
+    estimated_value TEXT NOT NULL,
+    actual_value TEXT,
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_estimation_category ON estimation_snapshots(category);
+CREATE INDEX IF NOT EXISTS idx_estimation_job ON estimation_snapshots(job_id);
+
+-- ==================== Self Repair ====================
+
+CREATE TABLE IF NOT EXISTS repair_attempts (
+    id TEXT PRIMARY KEY,
+    target_type TEXT NOT NULL,
+    target_id TEXT NOT NULL,
+    diagnosis TEXT NOT NULL,
+    action_taken TEXT NOT NULL,
+    success INTEGER NOT NULL,
+    error_message TEXT,
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_repair_attempts_target ON repair_attempts(target_type, target_id);
+CREATE INDEX IF NOT EXISTS idx_repair_attempts_created ON repair_attempts(created_at);
+
+-- ==================== Workspace: Memory Documents ====================
+
+CREATE TABLE IF NOT EXISTS memory_documents (
+    id TEXT PRIMARY KEY,
+    user_id TEXT NOT NULL,
+    agent_id TEXT,
+    path TEXT NOT NULL,
+    content TEXT NOT NULL,
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+    metadata TEXT NOT NULL DEFAULT '{}',
+    UNIQUE (user_id, agent_id, path)
+);
+
+CREATE INDEX IF NOT EXISTS idx_memory_documents_user ON memory_documents(user_id);
+CREATE INDEX IF NOT EXISTS idx_memory_documents_path ON memory_documents(user_id, path);
+CREATE INDEX IF NOT EXISTS idx_memory_documents_updated ON memory_documents(updated_at DESC);
+
+-- Trigger to auto-update updated_at on memory_documents
+CREATE TRIGGER IF NOT EXISTS update_memory_documents_updated_at
+    AFTER UPDATE ON memory_documents
+    FOR EACH ROW
+    WHEN NEW.updated_at = OLD.updated_at
+    BEGIN
+        UPDATE memory_documents SET updated_at = datetime('now') WHERE id = NEW.id;
+    END;
+
+-- ==================== Workspace: Memory Chunks ====================
+
+CREATE TABLE IF NOT EXISTS memory_chunks (
+    _rowid INTEGER PRIMARY KEY AUTOINCREMENT,
+    id TEXT NOT NULL UNIQUE,
+    document_id TEXT NOT NULL REFERENCES memory_documents(id) ON DELETE CASCADE,
+    chunk_index INTEGER NOT NULL,
+    content TEXT NOT NULL,
+    embedding F32_BLOB(1536),
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    UNIQUE (document_id, chunk_index)
+);
+
+CREATE INDEX IF NOT EXISTS idx_memory_chunks_document ON memory_chunks(document_id);
+
+-- Vector index for semantic search (libSQL native)
+CREATE INDEX IF NOT EXISTS idx_memory_chunks_embedding
+    ON memory_chunks (libsql_vector_idx(embedding));
+
+-- FTS5 virtual table for full-text search
+CREATE VIRTUAL TABLE IF NOT EXISTS memory_chunks_fts USING fts5(
+    content,
+    content='memory_chunks',
+    content_rowid='_rowid'
+);
+
+-- Triggers to keep FTS5 in sync with memory_chunks
+CREATE TRIGGER IF NOT EXISTS memory_chunks_fts_insert AFTER INSERT ON memory_chunks BEGIN
+    INSERT INTO memory_chunks_fts(rowid, content) VALUES (new._rowid, new.content);
+END;
+
+CREATE TRIGGER IF NOT EXISTS memory_chunks_fts_delete AFTER DELETE ON memory_chunks BEGIN
+    INSERT INTO memory_chunks_fts(memory_chunks_fts, rowid, content)
+        VALUES ('delete', old._rowid, old.content);
+END;
+
+CREATE TRIGGER IF NOT EXISTS memory_chunks_fts_update AFTER UPDATE ON memory_chunks BEGIN
+    INSERT INTO memory_chunks_fts(memory_chunks_fts, rowid, content)
+        VALUES ('delete', old._rowid, old.content);
+    INSERT INTO memory_chunks_fts(rowid, content) VALUES (new._rowid, new.content);
+END;
+
+-- ==================== Workspace: Heartbeat State ====================
+
+CREATE TABLE IF NOT EXISTS heartbeat_state (
+    id TEXT PRIMARY KEY,
+    user_id TEXT NOT NULL,
+    agent_id TEXT,
+    last_run TEXT,
+    next_run TEXT,
+    interval_seconds INTEGER NOT NULL DEFAULT 1800,
+    enabled INTEGER NOT NULL DEFAULT 1,
+    consecutive_failures INTEGER NOT NULL DEFAULT 0,
+    last_checks TEXT NOT NULL DEFAULT '{}',
+    UNIQUE (user_id, agent_id)
+);
+
+CREATE INDEX IF NOT EXISTS idx_heartbeat_user ON heartbeat_state(user_id);
+
+-- ==================== Secrets ====================
+
+CREATE TABLE IF NOT EXISTS secrets (
+    id TEXT PRIMARY KEY,
+    user_id TEXT NOT NULL,
+    name TEXT NOT NULL,
+    encrypted_value BLOB NOT NULL,
+    key_salt BLOB NOT NULL,
+    provider TEXT,
+    expires_at TEXT,
+    last_used_at TEXT,
+    usage_count INTEGER NOT NULL DEFAULT 0,
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+    UNIQUE (user_id, name)
+);
+
+CREATE INDEX IF NOT EXISTS idx_secrets_user ON secrets(user_id);
+
+-- ==================== WASM Tools ====================
+
+CREATE TABLE IF NOT EXISTS wasm_tools (
+    id TEXT PRIMARY KEY,
+    user_id TEXT NOT NULL,
+    name TEXT NOT NULL,
+    version TEXT NOT NULL DEFAULT '1.0.0',
+    description TEXT NOT NULL,
+    wasm_binary BLOB NOT NULL,
+    binary_hash BLOB NOT NULL,
+    parameters_schema TEXT NOT NULL,
+    source_url TEXT,
+    trust_level TEXT NOT NULL DEFAULT 'user',
+    status TEXT NOT NULL DEFAULT 'active',
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+    UNIQUE (user_id, name, version)
+);
+
+CREATE INDEX IF NOT EXISTS idx_wasm_tools_user ON wasm_tools(user_id);
+CREATE INDEX IF NOT EXISTS idx_wasm_tools_name ON wasm_tools(user_id, name);
+CREATE INDEX IF NOT EXISTS idx_wasm_tools_status ON wasm_tools(status);
+
+-- ==================== Tool Capabilities ====================
+
+CREATE TABLE IF NOT EXISTS tool_capabilities (
+    id TEXT PRIMARY KEY,
+    wasm_tool_id TEXT NOT NULL REFERENCES wasm_tools(id) ON DELETE CASCADE,
+    http_allowlist TEXT NOT NULL DEFAULT '[]',
+    allowed_secrets TEXT NOT NULL DEFAULT '[]',
+    tool_aliases TEXT NOT NULL DEFAULT '{}',
+    requests_per_minute INTEGER NOT NULL DEFAULT 60,
+    requests_per_hour INTEGER NOT NULL DEFAULT 1000,
+    max_request_body_bytes INTEGER NOT NULL DEFAULT 1048576,
+    max_response_body_bytes INTEGER NOT NULL DEFAULT 10485760,
+    workspace_read_prefixes TEXT NOT NULL DEFAULT '[]',
+    http_timeout_secs INTEGER NOT NULL DEFAULT 30,
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+    UNIQUE (wasm_tool_id)
+);
+
+-- ==================== Leak Detection Patterns ====================
+
+CREATE TABLE IF NOT EXISTS leak_detection_patterns (
+    id TEXT PRIMARY KEY,
+    name TEXT NOT NULL UNIQUE,
+    pattern TEXT NOT NULL,
+    severity TEXT NOT NULL DEFAULT 'high',
+    action TEXT NOT NULL DEFAULT 'block',
+    enabled INTEGER NOT NULL DEFAULT 1,
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+-- ==================== Rate Limit State ====================
+
+CREATE TABLE IF NOT EXISTS tool_rate_limit_state (
+    id TEXT PRIMARY KEY,
+    wasm_tool_id TEXT NOT NULL REFERENCES wasm_tools(id) ON DELETE CASCADE,
+    user_id TEXT NOT NULL,
+    minute_window_start TEXT NOT NULL DEFAULT (datetime('now')),
+    minute_count INTEGER NOT NULL DEFAULT 0,
+    hour_window_start TEXT NOT NULL DEFAULT (datetime('now')),
+    hour_count INTEGER NOT NULL DEFAULT 0,
+    UNIQUE (wasm_tool_id, user_id)
+);
+
+-- ==================== Secret Usage Audit Log ====================
+
+CREATE TABLE IF NOT EXISTS secret_usage_log (
+    id TEXT PRIMARY KEY,
+    secret_id TEXT NOT NULL REFERENCES secrets(id) ON DELETE CASCADE,
+    wasm_tool_id TEXT REFERENCES wasm_tools(id) ON DELETE SET NULL,
+    user_id TEXT NOT NULL,
+    target_host TEXT NOT NULL,
+    target_path TEXT,
+    success INTEGER NOT NULL,
+    error_message TEXT,
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_secret_usage_user ON secret_usage_log(user_id);
+
+-- ==================== Leak Detection Events ====================
+
+CREATE TABLE IF NOT EXISTS leak_detection_events (
+    id TEXT PRIMARY KEY,
+    pattern_id TEXT REFERENCES leak_detection_patterns(id) ON DELETE SET NULL,
+    wasm_tool_id TEXT REFERENCES wasm_tools(id) ON DELETE SET NULL,
+    user_id TEXT NOT NULL,
+    source TEXT NOT NULL,
+    action_taken TEXT NOT NULL,
+    context_preview TEXT,
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+-- ==================== Tool Failures ====================
+
+CREATE TABLE IF NOT EXISTS tool_failures (
+    id TEXT PRIMARY KEY,
+    tool_name TEXT NOT NULL UNIQUE,
+    error_message TEXT,
+    error_count INTEGER DEFAULT 1,
+    first_failure TEXT DEFAULT (datetime('now')),
+    last_failure TEXT DEFAULT (datetime('now')),
+    last_build_result TEXT,
+    repaired_at TEXT,
+    repair_attempts INTEGER DEFAULT 0
+);
+
+CREATE INDEX IF NOT EXISTS idx_tool_failures_name ON tool_failures(tool_name);
+
+-- ==================== Job Events ====================
+
+CREATE TABLE IF NOT EXISTS job_events (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    job_id TEXT NOT NULL REFERENCES agent_jobs(id),
+    event_type TEXT NOT NULL,
+    data TEXT NOT NULL,
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_job_events_job ON job_events(job_id, id);
+
+-- ==================== Routines ====================
+
+CREATE TABLE IF NOT EXISTS routines (
+    id TEXT PRIMARY KEY,
+    name TEXT NOT NULL,
+    description TEXT NOT NULL DEFAULT '',
+    user_id TEXT NOT NULL,
+    enabled INTEGER NOT NULL DEFAULT 1,
+    trigger_type TEXT NOT NULL,
+    trigger_config TEXT NOT NULL,
+    action_type TEXT NOT NULL,
+    action_config TEXT NOT NULL,
+    cooldown_secs INTEGER NOT NULL DEFAULT 300,
+    max_concurrent INTEGER NOT NULL DEFAULT 1,
+    dedup_window_secs INTEGER,
+    notify_channel TEXT,
+    notify_user TEXT NOT NULL DEFAULT 'default',
+    notify_on_success INTEGER NOT NULL DEFAULT 0,
+    notify_on_failure INTEGER NOT NULL DEFAULT 1,
+    notify_on_attention INTEGER NOT NULL DEFAULT 1,
+    state TEXT NOT NULL DEFAULT '{}',
+    last_run_at TEXT,
+    next_fire_at TEXT,
+    run_count INTEGER NOT NULL DEFAULT 0,
+    consecutive_failures INTEGER NOT NULL DEFAULT 0,
+    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+    updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+    UNIQUE (user_id, name)
+);
+
+CREATE INDEX IF NOT EXISTS idx_routines_user ON routines(user_id);
+
+-- ==================== Routine Runs ====================
+
+CREATE TABLE IF NOT EXISTS routine_runs (
+    id TEXT PRIMARY KEY,
+    routine_id TEXT NOT NULL REFERENCES routines(id) ON DELETE CASCADE,
+    trigger_type TEXT NOT NULL,
+    trigger_detail TEXT,
+    started_at TEXT NOT NULL DEFAULT (datetime('now')),
+    completed_at TEXT,
+    status TEXT NOT NULL DEFAULT 'running',
+    result_summary TEXT,
+    tokens_used INTEGER,
+    job_id TEXT REFERENCES agent_jobs(id),
+    created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_routine_runs_routine ON routine_runs(routine_id);
+
+-- ==================== Settings ====================
+
+CREATE TABLE IF NOT EXISTS settings (
+    user_id TEXT NOT NULL,
+    key TEXT NOT NULL,
+    value TEXT NOT NULL,
+    updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+    PRIMARY KEY (user_id, key)
+);
+
+CREATE INDEX IF NOT EXISTS idx_settings_user ON settings(user_id);
+
+-- ==================== Missing indexes (parity with PostgreSQL) ====================
+
+-- agent_jobs
+CREATE INDEX IF NOT EXISTS idx_agent_jobs_stuck ON agent_jobs(stuck_since);
+
+-- secrets
+CREATE INDEX IF NOT EXISTS idx_secrets_provider ON secrets(provider);
+CREATE INDEX IF NOT EXISTS idx_secrets_expires ON secrets(expires_at);
+
+-- wasm_tools
+CREATE INDEX IF NOT EXISTS idx_wasm_tools_trust ON wasm_tools(trust_level);
+
+-- tool_capabilities
+CREATE INDEX IF NOT EXISTS idx_tool_capabilities_tool ON tool_capabilities(wasm_tool_id);
+
+-- leak_detection_patterns
+CREATE INDEX IF NOT EXISTS idx_leak_patterns_enabled ON leak_detection_patterns(enabled);
+
+-- tool_rate_limit_state
+CREATE INDEX IF NOT EXISTS idx_rate_limit_tool ON tool_rate_limit_state(wasm_tool_id);
+
+-- secret_usage_log
+CREATE INDEX IF NOT EXISTS idx_secret_usage_secret ON secret_usage_log(secret_id);
+CREATE INDEX IF NOT EXISTS idx_secret_usage_tool ON secret_usage_log(wasm_tool_id);
+CREATE INDEX IF NOT EXISTS idx_secret_usage_created ON secret_usage_log(created_at DESC);
+
+-- leak_detection_events
+CREATE INDEX IF NOT EXISTS idx_leak_events_pattern ON leak_detection_events(pattern_id);
+CREATE INDEX IF NOT EXISTS idx_leak_events_tool ON leak_detection_events(wasm_tool_id);
+CREATE INDEX IF NOT EXISTS idx_leak_events_user ON leak_detection_events(user_id);
+CREATE INDEX IF NOT EXISTS idx_leak_events_created ON leak_detection_events(created_at DESC);
+
+-- tool_failures
+CREATE INDEX IF NOT EXISTS idx_tool_failures_count ON tool_failures(error_count DESC);
+CREATE INDEX IF NOT EXISTS idx_tool_failures_unrepaired ON tool_failures(tool_name);
+
+-- routines
+CREATE INDEX IF NOT EXISTS idx_routines_next_fire ON routines(next_fire_at);
+CREATE INDEX IF NOT EXISTS idx_routines_event_triggers ON routines(user_id);
+
+-- routine_runs
+CREATE INDEX IF NOT EXISTS idx_routine_runs_status ON routine_runs(status);
+
+-- heartbeat_state
+CREATE INDEX IF NOT EXISTS idx_heartbeat_next_run ON heartbeat_state(next_run);
+
+-- ==================== Seed data ====================
+
+-- Pre-populate leak detection patterns (matches PostgreSQL V2 migration).
+INSERT OR IGNORE INTO leak_detection_patterns (id, name, pattern, severity, action, enabled, created_at) VALUES
+    ('550e8400-e29b-41d4-a716-446655440001', 'openai_api_key', 'sk-(?:proj-)?[a-zA-Z0-9]{20,}(?:T3BlbkFJ[a-zA-Z0-9_-]*)?', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440002', 'anthropic_api_key', 'sk-ant-api[a-zA-Z0-9_-]{90,}', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440003', 'aws_access_key', 'AKIA[0-9A-Z]{16}', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440004', 'aws_secret_key', '(?<![A-Za-z0-9/+=])[A-Za-z0-9/+=]{40}(?![A-Za-z0-9/+=])', 'high', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440005', 'github_token', 'gh[pousr]_[A-Za-z0-9_]{36,}', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440006', 'github_fine_grained_pat', 'github_pat_[a-zA-Z0-9]{22}_[a-zA-Z0-9]{59}', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440007', 'stripe_api_key', 'sk_(?:live|test)_[a-zA-Z0-9]{24,}', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440008', 'nearai_session', 'sess_[a-zA-Z0-9]{32,}', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440009', 'bearer_token', 'Bearer\s+[a-zA-Z0-9_-]{20,}', 'high', 'redact', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-44665544000a', 'pem_private_key', '-----BEGIN\s+(?:RSA\s+)?PRIVATE\s+KEY-----', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-44665544000b', 'ssh_private_key', '-----BEGIN\s+(?:OPENSSH|EC|DSA)\s+PRIVATE\s+KEY-----', 'critical', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-44665544000c', 'google_api_key', 'AIza[0-9A-Za-z_-]{35}', 'high', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-44665544000d', 'slack_token', 'xox[baprs]-[0-9a-zA-Z-]{10,}', 'high', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-44665544000e', 'discord_token', '[MN][A-Za-z\d]{23,}\.[\w-]{6}\.[\w-]{27}', 'high', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-44665544000f', 'twilio_api_key', 'SK[a-fA-F0-9]{32}', 'high', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440010', 'sendgrid_api_key', 'SG\.[a-zA-Z0-9_-]{22}\.[a-zA-Z0-9_-]{43}', 'high', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440011', 'mailchimp_api_key', '[a-f0-9]{32}-us[0-9]{1,2}', 'medium', 'block', 1, datetime('now')),
+    ('550e8400-e29b-41d4-a716-446655440012', 'high_entropy_hex', '(?<![a-fA-F0-9])[a-fA-F0-9]{64}(?![a-fA-F0-9])', 'medium', 'warn', 1, datetime('now'));
+
+"#;

src/db/mod.rs

@@ -0,0 +1,538 @@
+//! Database abstraction layer.
+//!
+//! Provides a backend-agnostic `Database` trait that unifies all persistence
+//! operations. Two implementations exist behind feature flags:
+//!
+//! - `postgres` (default): Uses `deadpool-postgres` + `tokio-postgres`
+//! - `libsql`: Uses libSQL (Turso's SQLite fork) for embedded/edge deployment
+//!
+//! The existing `Store`, `Repository`, `SecretsStore`, and `WasmToolStore`
+//! types become thin wrappers that delegate to `Arc<dyn Database>`.
+
+#[cfg(feature = "postgres")]
+pub mod postgres;
+
+#[cfg(feature = "libsql")]
+pub mod libsql_backend;
+
+#[cfg(feature = "libsql")]
+pub mod libsql_migrations;
+
+use std::collections::HashMap;
+use std::sync::Arc;
+
+use async_trait::async_trait;
+use chrono::{DateTime, Utc};
+use rust_decimal::Decimal;
+use uuid::Uuid;
+
+use crate::agent::BrokenTool;
+use crate::agent::routine::{Routine, RoutineRun, RunStatus};
+use crate::context::{ActionRecord, JobContext, JobState};
+use crate::error::DatabaseError;
+use crate::error::WorkspaceError;
+use crate::history::{
+    ConversationMessage, ConversationSummary, JobEventRecord, LlmCallRecord, SandboxJobRecord,
+    SandboxJobSummary, SettingRow,
+};
+use crate::workspace::{MemoryChunk, MemoryDocument, WorkspaceEntry};
+use crate::workspace::{SearchConfig, SearchResult};
+
+/// Create a database backend from configuration, run migrations, and return it.
+///
+/// This is the shared helper for CLI commands and other call sites that need
+/// a simple `Arc<dyn Database>` without retaining backend-specific handles
+/// (e.g., `pg_pool` or `libsql_conn` for the secrets store). The main agent
+/// startup in `main.rs` uses its own initialization block because it also
+/// captures those backend-specific handles.
+pub async fn connect_from_config(
+    config: &crate::config::DatabaseConfig,
+) -> Result<Arc<dyn Database>, DatabaseError> {
+    match config.backend {
+        #[cfg(feature = "libsql")]
+        crate::config::DatabaseBackend::LibSql => {
+            use secrecy::ExposeSecret as _;
+
+            let default_path = crate::config::default_libsql_path();
+            let db_path = config.libsql_path.as_deref().unwrap_or(&default_path);
+
+            let backend = if let Some(ref url) = config.libsql_url {
+                let token = config.libsql_auth_token.as_ref().ok_or_else(|| {
+                    DatabaseError::Pool(
+                        "LIBSQL_AUTH_TOKEN required when LIBSQL_URL is set".to_string(),
+                    )
+                })?;
+                libsql_backend::LibSqlBackend::new_remote_replica(
+                    db_path,
+                    url,
+                    token.expose_secret(),
+                )
+                .await
+                .map_err(|e| DatabaseError::Pool(e.to_string()))?
+            } else {
+                libsql_backend::LibSqlBackend::new_local(db_path)
+                    .await
+                    .map_err(|e| DatabaseError::Pool(e.to_string()))?
+            };
+            backend.run_migrations().await?;
+            Ok(Arc::new(backend))
+        }
+        #[cfg(feature = "postgres")]
+        _ => {
+            let pg = postgres::PgBackend::new(config)
+                .await
+                .map_err(|e| DatabaseError::Pool(e.to_string()))?;
+            pg.run_migrations().await?;
+            Ok(Arc::new(pg))
+        }
+        #[cfg(not(feature = "postgres"))]
+        _ => Err(DatabaseError::Pool(
+            "No database backend available. Enable 'postgres' or 'libsql' feature.".to_string(),
+        )),
+    }
+}
+
+/// Backend-agnostic database trait.
+///
+/// Combines all persistence operations from Store, Repository, and related
+/// stores into a single trait that can be implemented for different backends.
+#[async_trait]
+pub trait Database: Send + Sync {
+    /// Run schema migrations for this backend.
+    async fn run_migrations(&self) -> Result<(), DatabaseError>;
+
+    // ==================== Conversations ====================
+
+    /// Create a new conversation.
+    async fn create_conversation(
+        &self,
+        channel: &str,
+        user_id: &str,
+        thread_id: Option<&str>,
+    ) -> Result<Uuid, DatabaseError>;
+
+    /// Update conversation last activity.
+    async fn touch_conversation(&self, id: Uuid) -> Result<(), DatabaseError>;
+
+    /// Add a message to a conversation.
+    async fn add_conversation_message(
+        &self,
+        conversation_id: Uuid,
+        role: &str,
+        content: &str,
+    ) -> Result<Uuid, DatabaseError>;
+
+    /// Ensure a conversation row exists (upsert).
+    async fn ensure_conversation(
+        &self,
+        id: Uuid,
+        channel: &str,
+        user_id: &str,
+        thread_id: Option<&str>,
+    ) -> Result<(), DatabaseError>;
+
+    /// List conversations with a title preview.
+    async fn list_conversations_with_preview(
+        &self,
+        user_id: &str,
+        channel: &str,
+        limit: i64,
+    ) -> Result<Vec<ConversationSummary>, DatabaseError>;
+
+    /// Get or create the singleton assistant conversation.
+    async fn get_or_create_assistant_conversation(
+        &self,
+        user_id: &str,
+        channel: &str,
+    ) -> Result<Uuid, DatabaseError>;
+
+    /// Create a conversation with specific metadata.
+    async fn create_conversation_with_metadata(
+        &self,
+        channel: &str,
+        user_id: &str,
+        metadata: &serde_json::Value,
+    ) -> Result<Uuid, DatabaseError>;
+
+    /// Load messages with cursor-based pagination.
+    async fn list_conversation_messages_paginated(
+        &self,
+        conversation_id: Uuid,
+        before: Option<DateTime<Utc>>,
+        limit: i64,
+    ) -> Result<(Vec<ConversationMessage>, bool), DatabaseError>;
+
+    /// Merge a single key into conversation metadata.
+    async fn update_conversation_metadata_field(
+        &self,
+        id: Uuid,
+        key: &str,
+        value: &serde_json::Value,
+    ) -> Result<(), DatabaseError>;
+
+    /// Read conversation metadata.
+    async fn get_conversation_metadata(
+        &self,
+        id: Uuid,
+    ) -> Result<Option<serde_json::Value>, DatabaseError>;
+
+    /// Load all messages for a conversation.
+    async fn list_conversation_messages(
+        &self,
+        conversation_id: Uuid,
+    ) -> Result<Vec<ConversationMessage>, DatabaseError>;
+
+    /// Check if a conversation belongs to a specific user.
+    async fn conversation_belongs_to_user(
+        &self,
+        conversation_id: Uuid,
+        user_id: &str,
+    ) -> Result<bool, DatabaseError>;
+
+    // ==================== Jobs ====================
+
+    /// Save a job context.
+    async fn save_job(&self, ctx: &JobContext) -> Result<(), DatabaseError>;
+
+    /// Get a job by ID.
+    async fn get_job(&self, id: Uuid) -> Result<Option<JobContext>, DatabaseError>;
+
+    /// Update job status.
+    async fn update_job_status(
+        &self,
+        id: Uuid,
+        status: JobState,
+        failure_reason: Option<&str>,
+    ) -> Result<(), DatabaseError>;
+
+    /// Mark job as stuck.
+    async fn mark_job_stuck(&self, id: Uuid) -> Result<(), DatabaseError>;
+
+    /// Get stuck jobs.
+    async fn get_stuck_jobs(&self) -> Result<Vec<Uuid>, DatabaseError>;
+
+    // ==================== Actions ====================
+
+    /// Save a job action.
+    async fn save_action(&self, job_id: Uuid, action: &ActionRecord) -> Result<(), DatabaseError>;
+
+    /// Get actions for a job.
+    async fn get_job_actions(&self, job_id: Uuid) -> Result<Vec<ActionRecord>, DatabaseError>;
+
+    // ==================== LLM Calls ====================
+
+    /// Record an LLM call.
+    async fn record_llm_call(&self, record: &LlmCallRecord<'_>) -> Result<Uuid, DatabaseError>;
+
+    // ==================== Estimation Snapshots ====================
+
+    /// Save an estimation snapshot.
+    async fn save_estimation_snapshot(
+        &self,
+        job_id: Uuid,
+        category: &str,
+        tool_names: &[String],
+        estimated_cost: Decimal,
+        estimated_time_secs: i32,
+        estimated_value: Decimal,
+    ) -> Result<Uuid, DatabaseError>;
+
+    /// Update estimation snapshot with actual values.
+    async fn update_estimation_actuals(
+        &self,
+        id: Uuid,
+        actual_cost: Decimal,
+        actual_time_secs: i32,
+        actual_value: Option<Decimal>,
+    ) -> Result<(), DatabaseError>;
+
+    // ==================== Sandbox Jobs ====================
+
+    /// Insert a new sandbox job.
+    async fn save_sandbox_job(&self, job: &SandboxJobRecord) -> Result<(), DatabaseError>;
+
+    /// Get a sandbox job by ID.
+    async fn get_sandbox_job(&self, id: Uuid) -> Result<Option<SandboxJobRecord>, DatabaseError>;
+
+    /// List all sandbox jobs, most recent first.
+    async fn list_sandbox_jobs(&self) -> Result<Vec<SandboxJobRecord>, DatabaseError>;
+
+    /// Update sandbox job status.
+    async fn update_sandbox_job_status(
+        &self,
+        id: Uuid,
+        status: &str,
+        success: Option<bool>,
+        message: Option<&str>,
+        started_at: Option<DateTime<Utc>>,
+        completed_at: Option<DateTime<Utc>>,
+    ) -> Result<(), DatabaseError>;
+
+    /// Mark stale sandbox jobs as interrupted.
+    async fn cleanup_stale_sandbox_jobs(&self) -> Result<u64, DatabaseError>;
+
+    /// Get sandbox job summary.
+    async fn sandbox_job_summary(&self) -> Result<SandboxJobSummary, DatabaseError>;
+
+    /// List sandbox jobs for a specific user, most recent first.
+    async fn list_sandbox_jobs_for_user(
+        &self,
+        user_id: &str,
+    ) -> Result<Vec<SandboxJobRecord>, DatabaseError>;
+
+    /// Get sandbox job summary for a specific user.
+    async fn sandbox_job_summary_for_user(
+        &self,
+        user_id: &str,
+    ) -> Result<SandboxJobSummary, DatabaseError>;
+
+    /// Check if a sandbox job belongs to a specific user.
+    async fn sandbox_job_belongs_to_user(
+        &self,
+        job_id: Uuid,
+        user_id: &str,
+    ) -> Result<bool, DatabaseError>;
+
+    /// Update sandbox job mode.
+    async fn update_sandbox_job_mode(&self, id: Uuid, mode: &str) -> Result<(), DatabaseError>;
+
+    /// Get sandbox job mode.
+    async fn get_sandbox_job_mode(&self, id: Uuid) -> Result<Option<String>, DatabaseError>;
+
+    // ==================== Job Events ====================
+
+    /// Persist a job event.
+    async fn save_job_event(
+        &self,
+        job_id: Uuid,
+        event_type: &str,
+        data: &serde_json::Value,
+    ) -> Result<(), DatabaseError>;
+
+    /// Load all job events.
+    async fn list_job_events(&self, job_id: Uuid) -> Result<Vec<JobEventRecord>, DatabaseError>;
+
+    // ==================== Routines ====================
+
+    /// Create a new routine.
+    async fn create_routine(&self, routine: &Routine) -> Result<(), DatabaseError>;
+
+    /// Get a routine by ID.
+    async fn get_routine(&self, id: Uuid) -> Result<Option<Routine>, DatabaseError>;
+
+    /// Get a routine by user_id and name.
+    async fn get_routine_by_name(
+        &self,
+        user_id: &str,
+        name: &str,
+    ) -> Result<Option<Routine>, DatabaseError>;
+
+    /// List routines for a user.
+    async fn list_routines(&self, user_id: &str) -> Result<Vec<Routine>, DatabaseError>;
+
+    /// List all enabled event routines.
+    async fn list_event_routines(&self) -> Result<Vec<Routine>, DatabaseError>;
+
+    /// List due cron routines.
+    async fn list_due_cron_routines(&self) -> Result<Vec<Routine>, DatabaseError>;
+
+    /// Update a routine.
+    async fn update_routine(&self, routine: &Routine) -> Result<(), DatabaseError>;
+
+    /// Update runtime state after a routine fires.
+    async fn update_routine_runtime(
+        &self,
+        id: Uuid,
+        last_run_at: DateTime<Utc>,
+        next_fire_at: Option<DateTime<Utc>>,
+        run_count: u64,
+        consecutive_failures: u32,
+        state: &serde_json::Value,
+    ) -> Result<(), DatabaseError>;
+
+    /// Delete a routine.
+    async fn delete_routine(&self, id: Uuid) -> Result<bool, DatabaseError>;
+
+    // ==================== Routine Runs ====================
+
+    /// Record a routine run starting.
+    async fn create_routine_run(&self, run: &RoutineRun) -> Result<(), DatabaseError>;
+
+    /// Complete a routine run.
+    async fn complete_routine_run(
+        &self,
+        id: Uuid,
+        status: RunStatus,
+        result_summary: Option<&str>,
+        tokens_used: Option<i32>,
+    ) -> Result<(), DatabaseError>;
+
+    /// List recent runs for a routine.
+    async fn list_routine_runs(
+        &self,
+        routine_id: Uuid,
+        limit: i64,
+    ) -> Result<Vec<RoutineRun>, DatabaseError>;
+
+    /// Count currently running runs for a routine.
+    async fn count_running_routine_runs(&self, routine_id: Uuid) -> Result<i64, DatabaseError>;
+
+    // ==================== Tool Failures ====================
+
+    /// Record a tool failure (upsert).
+    async fn record_tool_failure(
+        &self,
+        tool_name: &str,
+        error_message: &str,
+    ) -> Result<(), DatabaseError>;
+
+    /// Get broken tools exceeding threshold.
+    async fn get_broken_tools(&self, threshold: i32) -> Result<Vec<BrokenTool>, DatabaseError>;
+
+    /// Mark a tool as repaired.
+    async fn mark_tool_repaired(&self, tool_name: &str) -> Result<(), DatabaseError>;
+
+    /// Increment repair attempts.
+    async fn increment_repair_attempts(&self, tool_name: &str) -> Result<(), DatabaseError>;
+
+    // ==================== Settings ====================
+
+    /// Get a single setting.
+    async fn get_setting(
+        &self,
+        user_id: &str,
+        key: &str,
+    ) -> Result<Option<serde_json::Value>, DatabaseError>;
+
+    /// Get a single setting with metadata.
+    async fn get_setting_full(
+        &self,
+        user_id: &str,
+        key: &str,
+    ) -> Result<Option<SettingRow>, DatabaseError>;
+
+    /// Set a single setting (upsert).
+    async fn set_setting(
+        &self,
+        user_id: &str,
+        key: &str,
+        value: &serde_json::Value,
+    ) -> Result<(), DatabaseError>;
+
+    /// Delete a single setting.
+    async fn delete_setting(&self, user_id: &str, key: &str) -> Result<bool, DatabaseError>;
+
+    /// List all settings for a user.
+    async fn list_settings(&self, user_id: &str) -> Result<Vec<SettingRow>, DatabaseError>;
+
+    /// Get all settings as a flat map.
+    async fn get_all_settings(
+        &self,
+        user_id: &str,
+    ) -> Result<HashMap<String, serde_json::Value>, DatabaseError>;
+
+    /// Bulk-write settings atomically.
+    async fn set_all_settings(
+        &self,
+        user_id: &str,
+        settings: &HashMap<String, serde_json::Value>,
+    ) -> Result<(), DatabaseError>;
+
+    /// Check if settings exist for a user.
+    async fn has_settings(&self, user_id: &str) -> Result<bool, DatabaseError>;
+
+    // ==================== Workspace: Documents ====================
+
+    /// Get a document by path.
+    async fn get_document_by_path(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        path: &str,
+    ) -> Result<MemoryDocument, WorkspaceError>;
+
+    /// Get a document by ID.
+    async fn get_document_by_id(&self, id: Uuid) -> Result<MemoryDocument, WorkspaceError>;
+
+    /// Get or create a document by path.
+    async fn get_or_create_document_by_path(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        path: &str,
+    ) -> Result<MemoryDocument, WorkspaceError>;
+
+    /// Update a document's content.
+    async fn update_document(&self, id: Uuid, content: &str) -> Result<(), WorkspaceError>;
+
+    /// Delete a document by path.
+    async fn delete_document_by_path(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        path: &str,
+    ) -> Result<(), WorkspaceError>;
+
+    /// List files and directories in a directory path.
+    async fn list_directory(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        directory: &str,
+    ) -> Result<Vec<WorkspaceEntry>, WorkspaceError>;
+
+    /// List all file paths in the workspace.
+    async fn list_all_paths(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+    ) -> Result<Vec<String>, WorkspaceError>;
+
+    /// List all documents for a user.
+    async fn list_documents(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+    ) -> Result<Vec<MemoryDocument>, WorkspaceError>;
+
+    // ==================== Workspace: Chunks ====================
+
+    /// Delete all chunks for a document.
+    async fn delete_chunks(&self, document_id: Uuid) -> Result<(), WorkspaceError>;
+
+    /// Insert a chunk.
+    async fn insert_chunk(
+        &self,
+        document_id: Uuid,
+        chunk_index: i32,
+        content: &str,
+        embedding: Option<&[f32]>,
+    ) -> Result<Uuid, WorkspaceError>;
+
+    /// Update a chunk's embedding.
+    async fn update_chunk_embedding(
+        &self,
+        chunk_id: Uuid,
+        embedding: &[f32],
+    ) -> Result<(), WorkspaceError>;
+
+    /// Get chunks without embeddings for backfilling.
+    async fn get_chunks_without_embeddings(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        limit: usize,
+    ) -> Result<Vec<MemoryChunk>, WorkspaceError>;
+
+    // ==================== Workspace: Search ====================
+
+    /// Perform hybrid search combining FTS and vector similarity.
+    async fn hybrid_search(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        query: &str,
+        embedding: Option<&[f32]>,
+        config: &SearchConfig,
+    ) -> Result<Vec<SearchResult>, WorkspaceError>;
+}

src/db/postgres.rs

@@ -0,0 +1,627 @@
+//! PostgreSQL backend for the Database trait.
+//!
+//! Delegates to the existing `Store` (history) and `Repository` (workspace)
+//! implementations, avoiding SQL duplication.
+
+use std::collections::HashMap;
+
+use async_trait::async_trait;
+use chrono::{DateTime, Utc};
+use deadpool_postgres::Pool;
+use rust_decimal::Decimal;
+use uuid::Uuid;
+
+use crate::agent::BrokenTool;
+use crate::agent::routine::{Routine, RoutineRun, RunStatus};
+use crate::config::DatabaseConfig;
+use crate::context::{ActionRecord, JobContext, JobState};
+use crate::db::Database;
+use crate::error::{DatabaseError, WorkspaceError};
+use crate::history::{
+    ConversationMessage, ConversationSummary, JobEventRecord, LlmCallRecord, SandboxJobRecord,
+    SandboxJobSummary, SettingRow, Store,
+};
+use crate::workspace::{
+    MemoryChunk, MemoryDocument, Repository, SearchConfig, SearchResult, WorkspaceEntry,
+};
+
+/// PostgreSQL database backend.
+///
+/// Wraps the existing `Store` (for history/conversations/jobs/routines/settings)
+/// and `Repository` (for workspace documents/chunks/search) to implement the
+/// unified `Database` trait.
+pub struct PgBackend {
+    store: Store,
+    repo: Repository,
+}
+
+impl PgBackend {
+    /// Create a new PostgreSQL backend from configuration.
+    pub async fn new(config: &DatabaseConfig) -> Result<Self, DatabaseError> {
+        let store = Store::new(config).await?;
+        let repo = Repository::new(store.pool());
+        Ok(Self { store, repo })
+    }
+
+    /// Get a clone of the connection pool.
+    ///
+    /// Useful for sharing with components that still need raw pool access.
+    pub fn pool(&self) -> Pool {
+        self.store.pool()
+    }
+}
+
+#[async_trait]
+impl Database for PgBackend {
+    async fn run_migrations(&self) -> Result<(), DatabaseError> {
+        self.store.run_migrations().await
+    }
+
+    // ==================== Conversations ====================
+
+    async fn create_conversation(
+        &self,
+        channel: &str,
+        user_id: &str,
+        thread_id: Option<&str>,
+    ) -> Result<Uuid, DatabaseError> {
+        self.store
+            .create_conversation(channel, user_id, thread_id)
+            .await
+    }
+
+    async fn touch_conversation(&self, id: Uuid) -> Result<(), DatabaseError> {
+        self.store.touch_conversation(id).await
+    }
+
+    async fn add_conversation_message(
+        &self,
+        conversation_id: Uuid,
+        role: &str,
+        content: &str,
+    ) -> Result<Uuid, DatabaseError> {
+        self.store
+            .add_conversation_message(conversation_id, role, content)
+            .await
+    }
+
+    async fn ensure_conversation(
+        &self,
+        id: Uuid,
+        channel: &str,
+        user_id: &str,
+        thread_id: Option<&str>,
+    ) -> Result<(), DatabaseError> {
+        self.store
+            .ensure_conversation(id, channel, user_id, thread_id)
+            .await
+    }
+
+    async fn list_conversations_with_preview(
+        &self,
+        user_id: &str,
+        channel: &str,
+        limit: i64,
+    ) -> Result<Vec<ConversationSummary>, DatabaseError> {
+        self.store
+            .list_conversations_with_preview(user_id, channel, limit)
+            .await
+    }
+
+    async fn get_or_create_assistant_conversation(
+        &self,
+        user_id: &str,
+        channel: &str,
+    ) -> Result<Uuid, DatabaseError> {
+        self.store
+            .get_or_create_assistant_conversation(user_id, channel)
+            .await
+    }
+
+    async fn create_conversation_with_metadata(
+        &self,
+        channel: &str,
+        user_id: &str,
+        metadata: &serde_json::Value,
+    ) -> Result<Uuid, DatabaseError> {
+        self.store
+            .create_conversation_with_metadata(channel, user_id, metadata)
+            .await
+    }
+
+    async fn list_conversation_messages_paginated(
+        &self,
+        conversation_id: Uuid,
+        before: Option<DateTime<Utc>>,
+        limit: i64,
+    ) -> Result<(Vec<ConversationMessage>, bool), DatabaseError> {
+        self.store
+            .list_conversation_messages_paginated(conversation_id, before, limit)
+            .await
+    }
+
+    async fn update_conversation_metadata_field(
+        &self,
+        id: Uuid,
+        key: &str,
+        value: &serde_json::Value,
+    ) -> Result<(), DatabaseError> {
+        self.store
+            .update_conversation_metadata_field(id, key, value)
+            .await
+    }
+
+    async fn get_conversation_metadata(
+        &self,
+        id: Uuid,
+    ) -> Result<Option<serde_json::Value>, DatabaseError> {
+        self.store.get_conversation_metadata(id).await
+    }
+
+    async fn list_conversation_messages(
+        &self,
+        conversation_id: Uuid,
+    ) -> Result<Vec<ConversationMessage>, DatabaseError> {
+        self.store.list_conversation_messages(conversation_id).await
+    }
+
+    async fn conversation_belongs_to_user(
+        &self,
+        conversation_id: Uuid,
+        user_id: &str,
+    ) -> Result<bool, DatabaseError> {
+        self.store
+            .conversation_belongs_to_user(conversation_id, user_id)
+            .await
+    }
+
+    // ==================== Jobs ====================
+
+    async fn save_job(&self, ctx: &JobContext) -> Result<(), DatabaseError> {
+        self.store.save_job(ctx).await
+    }
+
+    async fn get_job(&self, id: Uuid) -> Result<Option<JobContext>, DatabaseError> {
+        self.store.get_job(id).await
+    }
+
+    async fn update_job_status(
+        &self,
+        id: Uuid,
+        status: JobState,
+        failure_reason: Option<&str>,
+    ) -> Result<(), DatabaseError> {
+        self.store
+            .update_job_status(id, status, failure_reason)
+            .await
+    }
+
+    async fn mark_job_stuck(&self, id: Uuid) -> Result<(), DatabaseError> {
+        self.store.mark_job_stuck(id).await
+    }
+
+    async fn get_stuck_jobs(&self) -> Result<Vec<Uuid>, DatabaseError> {
+        self.store.get_stuck_jobs().await
+    }
+
+    // ==================== Actions ====================
+
+    async fn save_action(&self, job_id: Uuid, action: &ActionRecord) -> Result<(), DatabaseError> {
+        self.store.save_action(job_id, action).await
+    }
+
+    async fn get_job_actions(&self, job_id: Uuid) -> Result<Vec<ActionRecord>, DatabaseError> {
+        self.store.get_job_actions(job_id).await
+    }
+
+    // ==================== LLM Calls ====================
+
+    async fn record_llm_call(&self, record: &LlmCallRecord<'_>) -> Result<Uuid, DatabaseError> {
+        self.store.record_llm_call(record).await
+    }
+
+    // ==================== Estimation Snapshots ====================
+
+    async fn save_estimation_snapshot(
+        &self,
+        job_id: Uuid,
+        category: &str,
+        tool_names: &[String],
+        estimated_cost: Decimal,
+        estimated_time_secs: i32,
+        estimated_value: Decimal,
+    ) -> Result<Uuid, DatabaseError> {
+        self.store
+            .save_estimation_snapshot(
+                job_id,
+                category,
+                tool_names,
+                estimated_cost,
+                estimated_time_secs,
+                estimated_value,
+            )
+            .await
+    }
+
+    async fn update_estimation_actuals(
+        &self,
+        id: Uuid,
+        actual_cost: Decimal,
+        actual_time_secs: i32,
+        actual_value: Option<Decimal>,
+    ) -> Result<(), DatabaseError> {
+        self.store
+            .update_estimation_actuals(id, actual_cost, actual_time_secs, actual_value)
+            .await
+    }
+
+    // ==================== Sandbox Jobs ====================
+
+    async fn save_sandbox_job(&self, job: &SandboxJobRecord) -> Result<(), DatabaseError> {
+        self.store.save_sandbox_job(job).await
+    }
+
+    async fn get_sandbox_job(&self, id: Uuid) -> Result<Option<SandboxJobRecord>, DatabaseError> {
+        self.store.get_sandbox_job(id).await
+    }
+
+    async fn list_sandbox_jobs(&self) -> Result<Vec<SandboxJobRecord>, DatabaseError> {
+        self.store.list_sandbox_jobs().await
+    }
+
+    async fn update_sandbox_job_status(
+        &self,
+        id: Uuid,
+        status: &str,
+        success: Option<bool>,
+        message: Option<&str>,
+        started_at: Option<DateTime<Utc>>,
+        completed_at: Option<DateTime<Utc>>,
+    ) -> Result<(), DatabaseError> {
+        self.store
+            .update_sandbox_job_status(id, status, success, message, started_at, completed_at)
+            .await
+    }
+
+    async fn cleanup_stale_sandbox_jobs(&self) -> Result<u64, DatabaseError> {
+        self.store.cleanup_stale_sandbox_jobs().await
+    }
+
+    async fn sandbox_job_summary(&self) -> Result<SandboxJobSummary, DatabaseError> {
+        self.store.sandbox_job_summary().await
+    }
+
+    async fn list_sandbox_jobs_for_user(
+        &self,
+        user_id: &str,
+    ) -> Result<Vec<SandboxJobRecord>, DatabaseError> {
+        self.store.list_sandbox_jobs_for_user(user_id).await
+    }
+
+    async fn sandbox_job_summary_for_user(
+        &self,
+        user_id: &str,
+    ) -> Result<SandboxJobSummary, DatabaseError> {
+        self.store.sandbox_job_summary_for_user(user_id).await
+    }
+
+    async fn sandbox_job_belongs_to_user(
+        &self,
+        job_id: Uuid,
+        user_id: &str,
+    ) -> Result<bool, DatabaseError> {
+        self.store
+            .sandbox_job_belongs_to_user(job_id, user_id)
+            .await
+    }
+
+    async fn update_sandbox_job_mode(&self, id: Uuid, mode: &str) -> Result<(), DatabaseError> {
+        self.store.update_sandbox_job_mode(id, mode).await
+    }
+
+    async fn get_sandbox_job_mode(&self, id: Uuid) -> Result<Option<String>, DatabaseError> {
+        self.store.get_sandbox_job_mode(id).await
+    }
+
+    // ==================== Job Events ====================
+
+    async fn save_job_event(
+        &self,
+        job_id: Uuid,
+        event_type: &str,
+        data: &serde_json::Value,
+    ) -> Result<(), DatabaseError> {
+        self.store.save_job_event(job_id, event_type, data).await
+    }
+
+    async fn list_job_events(&self, job_id: Uuid) -> Result<Vec<JobEventRecord>, DatabaseError> {
+        self.store.list_job_events(job_id).await
+    }
+
+    // ==================== Routines ====================
+
+    async fn create_routine(&self, routine: &Routine) -> Result<(), DatabaseError> {
+        self.store.create_routine(routine).await
+    }
+
+    async fn get_routine(&self, id: Uuid) -> Result<Option<Routine>, DatabaseError> {
+        self.store.get_routine(id).await
+    }
+
+    async fn get_routine_by_name(
+        &self,
+        user_id: &str,
+        name: &str,
+    ) -> Result<Option<Routine>, DatabaseError> {
+        self.store.get_routine_by_name(user_id, name).await
+    }
+
+    async fn list_routines(&self, user_id: &str) -> Result<Vec<Routine>, DatabaseError> {
+        self.store.list_routines(user_id).await
+    }
+
+    async fn list_event_routines(&self) -> Result<Vec<Routine>, DatabaseError> {
+        self.store.list_event_routines().await
+    }
+
+    async fn list_due_cron_routines(&self) -> Result<Vec<Routine>, DatabaseError> {
+        self.store.list_due_cron_routines().await
+    }
+
+    async fn update_routine(&self, routine: &Routine) -> Result<(), DatabaseError> {
+        self.store.update_routine(routine).await
+    }
+
+    async fn update_routine_runtime(
+        &self,
+        id: Uuid,
+        last_run_at: DateTime<Utc>,
+        next_fire_at: Option<DateTime<Utc>>,
+        run_count: u64,
+        consecutive_failures: u32,
+        state: &serde_json::Value,
+    ) -> Result<(), DatabaseError> {
+        self.store
+            .update_routine_runtime(
+                id,
+                last_run_at,
+                next_fire_at,
+                run_count,
+                consecutive_failures,
+                state,
+            )
+            .await
+    }
+
+    async fn delete_routine(&self, id: Uuid) -> Result<bool, DatabaseError> {
+        self.store.delete_routine(id).await
+    }
+
+    // ==================== Routine Runs ====================
+
+    async fn create_routine_run(&self, run: &RoutineRun) -> Result<(), DatabaseError> {
+        self.store.create_routine_run(run).await
+    }
+
+    async fn complete_routine_run(
+        &self,
+        id: Uuid,
+        status: RunStatus,
+        result_summary: Option<&str>,
+        tokens_used: Option<i32>,
+    ) -> Result<(), DatabaseError> {
+        self.store
+            .complete_routine_run(id, status, result_summary, tokens_used)
+            .await
+    }
+
+    async fn list_routine_runs(
+        &self,
+        routine_id: Uuid,
+        limit: i64,
+    ) -> Result<Vec<RoutineRun>, DatabaseError> {
+        self.store.list_routine_runs(routine_id, limit).await
+    }
+
+    async fn count_running_routine_runs(&self, routine_id: Uuid) -> Result<i64, DatabaseError> {
+        self.store.count_running_routine_runs(routine_id).await
+    }
+
+    // ==================== Tool Failures ====================
+
+    async fn record_tool_failure(
+        &self,
+        tool_name: &str,
+        error_message: &str,
+    ) -> Result<(), DatabaseError> {
+        self.store
+            .record_tool_failure(tool_name, error_message)
+            .await
+    }
+
+    async fn get_broken_tools(&self, threshold: i32) -> Result<Vec<BrokenTool>, DatabaseError> {
+        self.store.get_broken_tools(threshold).await
+    }
+
+    async fn mark_tool_repaired(&self, tool_name: &str) -> Result<(), DatabaseError> {
+        self.store.mark_tool_repaired(tool_name).await
+    }
+
+    async fn increment_repair_attempts(&self, tool_name: &str) -> Result<(), DatabaseError> {
+        self.store.increment_repair_attempts(tool_name).await
+    }
+
+    // ==================== Settings ====================
+
+    async fn get_setting(
+        &self,
+        user_id: &str,
+        key: &str,
+    ) -> Result<Option<serde_json::Value>, DatabaseError> {
+        self.store.get_setting(user_id, key).await
+    }
+
+    async fn get_setting_full(
+        &self,
+        user_id: &str,
+        key: &str,
+    ) -> Result<Option<SettingRow>, DatabaseError> {
+        self.store.get_setting_full(user_id, key).await
+    }
+
+    async fn set_setting(
+        &self,
+        user_id: &str,
+        key: &str,
+        value: &serde_json::Value,
+    ) -> Result<(), DatabaseError> {
+        self.store.set_setting(user_id, key, value).await
+    }
+
+    async fn delete_setting(&self, user_id: &str, key: &str) -> Result<bool, DatabaseError> {
+        self.store.delete_setting(user_id, key).await
+    }
+
+    async fn list_settings(&self, user_id: &str) -> Result<Vec<SettingRow>, DatabaseError> {
+        self.store.list_settings(user_id).await
+    }
+
+    async fn get_all_settings(
+        &self,
+        user_id: &str,
+    ) -> Result<HashMap<String, serde_json::Value>, DatabaseError> {
+        self.store.get_all_settings(user_id).await
+    }
+
+    async fn set_all_settings(
+        &self,
+        user_id: &str,
+        settings: &HashMap<String, serde_json::Value>,
+    ) -> Result<(), DatabaseError> {
+        self.store.set_all_settings(user_id, settings).await
+    }
+
+    async fn has_settings(&self, user_id: &str) -> Result<bool, DatabaseError> {
+        self.store.has_settings(user_id).await
+    }
+
+    // ==================== Workspace: Documents ====================
+
+    async fn get_document_by_path(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        path: &str,
+    ) -> Result<MemoryDocument, WorkspaceError> {
+        self.repo
+            .get_document_by_path(user_id, agent_id, path)
+            .await
+    }
+
+    async fn get_document_by_id(&self, id: Uuid) -> Result<MemoryDocument, WorkspaceError> {
+        self.repo.get_document_by_id(id).await
+    }
+
+    async fn get_or_create_document_by_path(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        path: &str,
+    ) -> Result<MemoryDocument, WorkspaceError> {
+        self.repo
+            .get_or_create_document_by_path(user_id, agent_id, path)
+            .await
+    }
+
+    async fn update_document(&self, id: Uuid, content: &str) -> Result<(), WorkspaceError> {
+        self.repo.update_document(id, content).await
+    }
+
+    async fn delete_document_by_path(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        path: &str,
+    ) -> Result<(), WorkspaceError> {
+        self.repo
+            .delete_document_by_path(user_id, agent_id, path)
+            .await
+    }
+
+    async fn list_directory(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        directory: &str,
+    ) -> Result<Vec<WorkspaceEntry>, WorkspaceError> {
+        self.repo.list_directory(user_id, agent_id, directory).await
+    }
+
+    async fn list_all_paths(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+    ) -> Result<Vec<String>, WorkspaceError> {
+        self.repo.list_all_paths(user_id, agent_id).await
+    }
+
+    async fn list_documents(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+    ) -> Result<Vec<MemoryDocument>, WorkspaceError> {
+        self.repo.list_documents(user_id, agent_id).await
+    }
+
+    // ==================== Workspace: Chunks ====================
+
+    async fn delete_chunks(&self, document_id: Uuid) -> Result<(), WorkspaceError> {
+        self.repo.delete_chunks(document_id).await
+    }
+
+    async fn insert_chunk(
+        &self,
+        document_id: Uuid,
+        chunk_index: i32,
+        content: &str,
+        embedding: Option<&[f32]>,
+    ) -> Result<Uuid, WorkspaceError> {
+        self.repo
+            .insert_chunk(document_id, chunk_index, content, embedding)
+            .await
+    }
+
+    async fn update_chunk_embedding(
+        &self,
+        chunk_id: Uuid,
+        embedding: &[f32],
+    ) -> Result<(), WorkspaceError> {
+        self.repo.update_chunk_embedding(chunk_id, embedding).await
+    }
+
+    async fn get_chunks_without_embeddings(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        limit: usize,
+    ) -> Result<Vec<MemoryChunk>, WorkspaceError> {
+        self.repo
+            .get_chunks_without_embeddings(user_id, agent_id, limit)
+            .await
+    }
+
+    // ==================== Workspace: Search ====================
+
+    async fn hybrid_search(
+        &self,
+        user_id: &str,
+        agent_id: Option<Uuid>,
+        query: &str,
+        embedding: Option<&[f32]>,
+        config: &SearchConfig,
+    ) -> Result<Vec<SearchResult>, WorkspaceError> {
+        self.repo
+            .hybrid_search(user_id, agent_id, query, embedding, config)
+            .await
+    }
+}

src/error.rs

@@ -40,6 +40,9 @@ pub enum Error {
     #[error("Workspace error: {0}")]
     Workspace(#[from] WorkspaceError),
 
+    #[error("Hook error: {0}")]
+    Hook(#[from] crate::hooks::HookError),
+
     #[error("Orchestrator error: {0}")]
     Orchestrator(#[from] OrchestratorError),
 
@@ -87,14 +90,21 @@ pub enum DatabaseError {
     #[error("Serialization error: {0}")]
     Serialization(String),
 
+    #[cfg(feature = "postgres")]
     #[error("PostgreSQL error: {0}")]
     Postgres(#[from] tokio_postgres::Error),
 
+    #[cfg(feature = "postgres")]
     #[error("Pool build error: {0}")]
     PoolBuild(#[from] deadpool_postgres::BuildError),
 
+    #[cfg(feature = "postgres")]
     #[error("Pool runtime error: {0}")]
     PoolRuntime(#[from] deadpool_postgres::PoolError),
+
+    #[cfg(feature = "libsql")]
+    #[error("LibSQL error: {0}")]
+    LibSql(#[from] libsql::Error),
 }
 
 /// Channel-related errors.

src/estimation/cost.rs

@@ -20,10 +20,6 @@ impl CostEstimator {
 
         // Default tool costs (in USD or equivalent)
         tool_costs.insert("http".to_string(), dec!(0.0001)); // API call
-        tool_costs.insert("marketplace".to_string(), dec!(0.01)); // Gas costs
-        tool_costs.insert("ecommerce".to_string(), dec!(0.001)); // API call
-        tool_costs.insert("taskrabbit".to_string(), dec!(0.0)); // Cost comes from task itself
-        tool_costs.insert("restaurant".to_string(), dec!(0.001)); // API call
         tool_costs.insert("echo".to_string(), dec!(0.0)); // Free
         tool_costs.insert("time".to_string(), dec!(0.0)); // Free
         tool_costs.insert("json".to_string(), dec!(0.0)); // Free
@@ -74,7 +70,7 @@ mod tests {
         let estimator = CostEstimator::new();
 
         assert_eq!(estimator.estimate_tool("echo"), dec!(0.0));
-        assert_eq!(estimator.estimate_tool("marketplace"), dec!(0.01));
+        assert_eq!(estimator.estimate_tool("http"), dec!(0.0001));
         assert!(estimator.estimate_tool("unknown") > dec!(0.0));
     }
 

src/estimation/time.rs

@@ -16,10 +16,6 @@ impl TimeEstimator {
 
         // Default tool durations
         tool_durations.insert("http".to_string(), Duration::from_secs(5));
-        tool_durations.insert("marketplace".to_string(), Duration::from_secs(10));
-        tool_durations.insert("ecommerce".to_string(), Duration::from_secs(8));
-        tool_durations.insert("taskrabbit".to_string(), Duration::from_secs(30)); // Just API, not task itself
-        tool_durations.insert("restaurant".to_string(), Duration::from_secs(5));
         tool_durations.insert("echo".to_string(), Duration::from_millis(10));
         tool_durations.insert("time".to_string(), Duration::from_millis(1));
         tool_durations.insert("json".to_string(), Duration::from_millis(5));

src/evaluation/success.rs

@@ -144,12 +144,11 @@ impl SuccessEvaluator for RuleBasedEvaluator {
 
         // Check for critical errors
         for action in actions.iter().filter(|a| !a.success) {
-            if let Some(ref error) = action.error {
-                if error.to_lowercase().contains("critical")
-                    || error.to_lowercase().contains("fatal")
-                {
-                    issues.push(format!("Critical error in {}: {}", action.tool_name, error));
-                }
+            if let Some(ref error) = action.error
+                && (error.to_lowercase().contains("critical")
+                    || error.to_lowercase().contains("fatal"))
+            {
+                issues.push(format!("Critical error in {}: {}", action.tool_name, error));
             }
         }
 

src/extensions/manager.rs

@@ -57,7 +57,7 @@ pub struct ExtensionManager {
     _tunnel_url: Option<String>,
     user_id: String,
     /// Optional database store for DB-backed MCP config.
-    store: Option<Arc<crate::history::Store>>,
+    store: Option<Arc<dyn crate::db::Database>>,
 }
 
 impl ExtensionManager {
@@ -71,7 +71,7 @@ impl ExtensionManager {
         wasm_channels_dir: PathBuf,
         tunnel_url: Option<String>,
         user_id: String,
-        store: Option<Arc<crate::history::Store>>,
+        store: Option<Arc<dyn crate::db::Database>>,
     ) -> Self {
         Self {
             registry: ExtensionRegistry::new(),
@@ -351,7 +351,7 @@ impl ExtensionManager {
     ) -> Result<crate::tools::mcp::config::McpServersFile, crate::tools::mcp::config::ConfigError>
     {
         if let Some(ref store) = self.store {
-            crate::tools::mcp::config::load_mcp_servers_from_db(store, &self.user_id).await
+            crate::tools::mcp::config::load_mcp_servers_from_db(store.as_ref(), &self.user_id).await
         } else {
             crate::tools::mcp::config::load_mcp_servers().await
         }
@@ -375,7 +375,8 @@ impl ExtensionManager {
     ) -> Result<(), crate::tools::mcp::config::ConfigError> {
         config.validate()?;
         if let Some(ref store) = self.store {
-            crate::tools::mcp::config::add_mcp_server_db(store, &self.user_id, config).await
+            crate::tools::mcp::config::add_mcp_server_db(store.as_ref(), &self.user_id, config)
+                .await
         } else {
             crate::tools::mcp::config::add_mcp_server(config).await
         }
@@ -386,7 +387,8 @@ impl ExtensionManager {
         name: &str,
     ) -> Result<(), crate::tools::mcp::config::ConfigError> {
         if let Some(ref store) = self.store {
-            crate::tools::mcp::config::remove_mcp_server_db(store, &self.user_id, name).await
+            crate::tools::mcp::config::remove_mcp_server_db(store.as_ref(), &self.user_id, name)
+                .await
         } else {
             crate::tools::mcp::config::remove_mcp_server(name).await
         }
@@ -461,7 +463,16 @@ impl ExtensionManager {
         name: &str,
         url: &str,
     ) -> Result<InstallResult, ExtensionError> {
-        // Download the WASM binary
+        // Require HTTPS to prevent downgrade attacks
+        if !url.starts_with("https://") {
+            return Err(ExtensionError::InstallFailed(
+                "Only HTTPS URLs are allowed for extension downloads".to_string(),
+            ));
+        }
+
+        // 50 MB cap to prevent disk-fill DoS
+        const MAX_WASM_SIZE: usize = 50 * 1024 * 1024;
+
         let client = reqwest::Client::builder()
             .timeout(std::time::Duration::from_secs(60))
             .build()
@@ -480,11 +491,36 @@ impl ExtensionManager {
             )));
         }
 
+        // Check Content-Length header before downloading the full body
+        if let Some(len) = response.content_length()
+            && len as usize > MAX_WASM_SIZE
+        {
+            return Err(ExtensionError::InstallFailed(format!(
+                "WASM binary too large ({} bytes, max {} bytes)",
+                len, MAX_WASM_SIZE
+            )));
+        }
+
         let bytes = response
             .bytes()
             .await
             .map_err(|e| ExtensionError::DownloadFailed(e.to_string()))?;
 
+        if bytes.len() > MAX_WASM_SIZE {
+            return Err(ExtensionError::InstallFailed(format!(
+                "WASM binary too large ({} bytes, max {} bytes)",
+                bytes.len(),
+                MAX_WASM_SIZE
+            )));
+        }
+
+        // Basic WASM magic number check (\0asm)
+        if bytes.len() < 4 || &bytes[..4] != b"\0asm" {
+            return Err(ExtensionError::InstallFailed(
+                "Downloaded file is not a valid WASM binary (bad magic number)".to_string(),
+            ));
+        }
+
         // Ensure tools directory exists
         tokio::fs::create_dir_all(&self.wasm_tools_dir)
             .await
@@ -497,9 +533,10 @@ impl ExtensionManager {
             .map_err(|e| ExtensionError::InstallFailed(e.to_string()))?;
 
         tracing::info!(
-            "Installed WASM tool '{}' ({} bytes) to {}",
+            "Installed WASM tool '{}' ({} bytes) from {} to {}",
             name,
             bytes.len(),
+            url,
             wasm_path.display()
         );
 
@@ -731,27 +768,27 @@ impl ExtensionManager {
         };
 
         // Check env var first
-        if let Some(ref env_var) = auth.env_var {
-            if let Ok(value) = std::env::var(env_var) {
-                // Store the env var value as a secret
-                let params = CreateSecretParams::new(&auth.secret_name, &value)
-                    .with_provider(name.to_string());
-                self.secrets
-                    .create(&self.user_id, params)
-                    .await
-                    .map_err(|e| ExtensionError::AuthFailed(e.to_string()))?;
+        if let Some(ref env_var) = auth.env_var
+            && let Ok(value) = std::env::var(env_var)
+        {
+            // Store the env var value as a secret
+            let params =
+                CreateSecretParams::new(&auth.secret_name, &value).with_provider(name.to_string());
+            self.secrets
+                .create(&self.user_id, params)
+                .await
+                .map_err(|e| ExtensionError::AuthFailed(e.to_string()))?;
 
-                return Ok(AuthResult {
-                    name: name.to_string(),
-                    kind: ExtensionKind::WasmTool,
-                    auth_url: None,
-                    callback_type: None,
-                    instructions: None,
-                    setup_url: None,
-                    awaiting_token: false,
-                    status: "authenticated".to_string(),
-                });
-            }
+            return Ok(AuthResult {
+                name: name.to_string(),
+                kind: ExtensionKind::WasmTool,
+                auth_url: None,
+                callback_type: None,
+                instructions: None,
+                setup_url: None,
+                awaiting_token: false,
+                status: "authenticated".to_string(),
+            });
         }
 
         // Check if already authenticated

src/history/mod.rs

@@ -5,11 +5,15 @@
 //! - Learning from past executions
 //! - Analytics and metrics
 
+#[cfg(feature = "postgres")]
 mod analytics;
 mod store;
 
+#[cfg(feature = "postgres")]
 pub use analytics::{JobStats, ToolStats};
+#[cfg(feature = "postgres")]
+pub use store::Store;
 pub use store::{
     ConversationMessage, ConversationSummary, JobEventRecord, LlmCallRecord, SandboxJobRecord,
-    SandboxJobSummary, Store,
+    SandboxJobSummary, SettingRow,
 };

src/history/store.rs

@@ -1,13 +1,18 @@
 //! PostgreSQL store for persisting agent data.
 
 use chrono::{DateTime, Utc};
+#[cfg(feature = "postgres")]
 use deadpool_postgres::{Config, Pool, Runtime};
 use rust_decimal::Decimal;
+#[cfg(feature = "postgres")]
 use tokio_postgres::NoTls;
 use uuid::Uuid;
 
+#[cfg(feature = "postgres")]
 use crate::config::DatabaseConfig;
+#[cfg(feature = "postgres")]
 use crate::context::{ActionRecord, JobContext, JobState};
+#[cfg(feature = "postgres")]
 use crate::error::DatabaseError;
 
 /// Record for an LLM call to be persisted.
@@ -24,11 +29,18 @@ pub struct LlmCallRecord<'a> {
 }
 
 /// Database store for the agent.
+#[cfg(feature = "postgres")]
 pub struct Store {
     pool: Pool,
 }
 
+#[cfg(feature = "postgres")]
 impl Store {
+    /// Wrap an existing pool (useful when the caller already has a connection).
+    pub fn from_pool(pool: Pool) -> Self {
+        Self { pool }
+    }
+
     /// Create a new store and connect to the database.
     pub async fn new(config: &DatabaseConfig) -> Result<Self, DatabaseError> {
         let mut cfg = Config::new();
@@ -144,7 +156,12 @@ impl Store {
                 actual_cost, repair_attempts, created_at, started_at, completed_at
             ) VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11, $12, $13, $14, $15, $16, $17)
             ON CONFLICT (id) DO UPDATE SET
+                title = EXCLUDED.title,
+                description = EXCLUDED.description,
+                category = EXCLUDED.category,
                 status = EXCLUDED.status,
+                estimated_cost = EXCLUDED.estimated_cost,
+                estimated_time_secs = EXCLUDED.estimated_time_secs,
                 actual_cost = EXCLUDED.actual_cost,
                 repair_attempts = EXCLUDED.repair_attempts,
                 started_at = EXCLUDED.started_at,
@@ -220,6 +237,8 @@ impl Store {
                     completed_at: row.get("completed_at"),
                     transitions: Vec::new(), // Not loaded from DB for now
                     metadata: serde_json::Value::Null,
+                    total_tokens_used: 0,
+                    max_tokens: 0,
                 }))
             }
             None => Ok(None),
@@ -464,6 +483,7 @@ pub struct SandboxJobSummary {
     pub interrupted: usize,
 }
 
+#[cfg(feature = "postgres")]
 impl Store {
     /// Insert a new sandbox job into `agent_jobs`.
     pub async fn save_sandbox_job(&self, job: &SandboxJobRecord) -> Result<(), DatabaseError> {
@@ -565,6 +585,90 @@ impl Store {
             .collect())
     }
 
+    /// List sandbox jobs for a specific user, most recent first.
+    pub async fn list_sandbox_jobs_for_user(
+        &self,
+        user_id: &str,
+    ) -> Result<Vec<SandboxJobRecord>, DatabaseError> {
+        let conn = self.conn().await?;
+        let rows = conn
+            .query(
+                r#"
+                SELECT id, title, status, user_id, project_dir,
+                       success, failure_reason, created_at, started_at, completed_at
+                FROM agent_jobs WHERE source = 'sandbox' AND user_id = $1
+                ORDER BY created_at DESC
+                "#,
+                &[&user_id],
+            )
+            .await?;
+
+        Ok(rows
+            .iter()
+            .map(|r| SandboxJobRecord {
+                id: r.get("id"),
+                task: r.get("title"),
+                status: r.get("status"),
+                user_id: r.get("user_id"),
+                project_dir: r
+                    .get::<_, Option<String>>("project_dir")
+                    .unwrap_or_default(),
+                success: r.get("success"),
+                failure_reason: r.get("failure_reason"),
+                created_at: r.get("created_at"),
+                started_at: r.get("started_at"),
+                completed_at: r.get("completed_at"),
+            })
+            .collect())
+    }
+
+    /// Get a summary of sandbox job counts by status for a specific user.
+    pub async fn sandbox_job_summary_for_user(
+        &self,
+        user_id: &str,
+    ) -> Result<SandboxJobSummary, DatabaseError> {
+        let conn = self.conn().await?;
+        let rows = conn
+            .query(
+                "SELECT status, COUNT(*) as cnt FROM agent_jobs WHERE source = 'sandbox' AND user_id = $1 GROUP BY status",
+                &[&user_id],
+            )
+            .await?;
+
+        let mut summary = SandboxJobSummary::default();
+        for row in &rows {
+            let status: String = row.get("status");
+            let count: i64 = row.get("cnt");
+            let c = count as usize;
+            summary.total += c;
+            match status.as_str() {
+                "creating" => summary.creating += c,
+                "running" => summary.running += c,
+                "completed" => summary.completed += c,
+                "failed" => summary.failed += c,
+                "interrupted" => summary.interrupted += c,
+                _ => {}
+            }
+        }
+        Ok(summary)
+    }
+
+    /// Check if a sandbox job belongs to a specific user.
+    pub async fn sandbox_job_belongs_to_user(
+        &self,
+        job_id: Uuid,
+        user_id: &str,
+    ) -> Result<bool, DatabaseError> {
+        let conn = self.conn().await?;
+        let row = conn
+            .query_opt(
+                "SELECT 1 FROM agent_jobs WHERE id = $1 AND user_id = $2 AND source = 'sandbox'",
+                &[&job_id, &user_id],
+            )
+            .await?;
+        Ok(row.is_some())
+    }
+
     /// Update sandbox job status and optional timestamps/result.
     pub async fn update_sandbox_job_status(
         &self,
@@ -656,6 +760,7 @@ pub struct JobEventRecord {
     pub created_at: DateTime<Utc>,
 }
 
+#[cfg(feature = "postgres")]
 impl Store {
     /// Persist a job event (fire-and-forget from orchestrator handler).
     pub async fn save_job_event(
@@ -728,10 +833,12 @@ impl Store {
 
 // ==================== Routines ====================
 
+#[cfg(feature = "postgres")]
 use crate::agent::routine::{
     NotifyConfig, Routine, RoutineAction, RoutineGuardrails, RoutineRun, RunStatus, Trigger,
 };
 
+#[cfg(feature = "postgres")]
 impl Store {
     /// Create a new routine.
     pub async fn create_routine(&self, routine: &Routine) -> Result<(), DatabaseError> {
@@ -1032,6 +1139,7 @@ impl Store {
     }
 }
 
+#[cfg(feature = "postgres")]
 fn row_to_routine(row: &tokio_postgres::Row) -> Result<Routine, DatabaseError> {
     let trigger_type: String = row.get("trigger_type");
     let trigger_config: serde_json::Value = row.get("trigger_config");
@@ -1076,6 +1184,7 @@ fn row_to_routine(row: &tokio_postgres::Row) -> Result<Routine, DatabaseError> {
     })
 }
 
+#[cfg(feature = "postgres")]
 fn row_to_routine_run(row: &tokio_postgres::Row) -> Result<RoutineRun, DatabaseError> {
     let status_str: String = row.get("status");
     let status: RunStatus = status_str
@@ -1121,6 +1230,7 @@ pub struct ConversationMessage {
     pub created_at: DateTime<Utc>,
 }
 
+#[cfg(feature = "postgres")]
 impl Store {
     /// Ensure a conversation row exists for a given UUID.
     ///
@@ -1258,6 +1368,22 @@ impl Store {
         Ok(id)
     }
 
+    /// Check whether a conversation belongs to the given user.
+    pub async fn conversation_belongs_to_user(
+        &self,
+        conversation_id: Uuid,
+        user_id: &str,
+    ) -> Result<bool, DatabaseError> {
+        let conn = self.conn().await?;
+        let row = conn
+            .query_opt(
+                "SELECT 1 FROM conversations WHERE id = $1 AND user_id = $2",
+                &[&conversation_id, &user_id],
+            )
+            .await?;
+        Ok(row.is_some())
+    }
+
     /// Load messages for a conversation with cursor-based pagination.
     ///
     /// Returns `(messages_oldest_first, has_more)`.
@@ -1375,6 +1501,7 @@ impl Store {
     }
 }
 
+#[cfg(feature = "postgres")]
 fn parse_job_state(s: &str) -> JobState {
     match s {
         "pending" => JobState::Pending,
@@ -1391,8 +1518,10 @@ fn parse_job_state(s: &str) -> JobState {
 
 // ==================== Tool Failures ====================
 
+#[cfg(feature = "postgres")]
 use crate::agent::BrokenTool;
 
+#[cfg(feature = "postgres")]
 impl Store {
     /// Record a tool failure (upsert: increment count if exists).
     pub async fn record_tool_failure(
@@ -1486,6 +1615,7 @@ pub struct SettingRow {
     pub updated_at: DateTime<Utc>,
 }
 
+#[cfg(feature = "postgres")]
 impl Store {
     /// Get a single setting by key.
     pub async fn get_setting(

src/hooks/hook.rs

@@ -0,0 +1,199 @@
+//! Core hook types and traits.
+
+use std::time::Duration;
+
+use async_trait::async_trait;
+
+/// Points in the agent lifecycle where hooks can be attached.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum HookPoint {
+    /// Before processing an inbound user message.
+    BeforeInbound,
+    /// Before executing a tool call.
+    BeforeToolCall,
+    /// Before sending an outbound response.
+    BeforeOutbound,
+    /// When a new session starts.
+    OnSessionStart,
+    /// When a session ends (pruned or expired).
+    OnSessionEnd,
+    /// Transform the final response before completing a turn.
+    TransformResponse,
+}
+
+/// Contextual data carried with each hook invocation.
+#[derive(Debug, Clone)]
+pub enum HookEvent {
+    /// An inbound user message about to be processed.
+    Inbound {
+        user_id: String,
+        channel: String,
+        content: String,
+        thread_id: Option<String>,
+    },
+    /// A tool call about to be executed.
+    ToolCall {
+        tool_name: String,
+        parameters: serde_json::Value,
+        user_id: String,
+        /// "chat" for interactive, or a job ID string for autonomous jobs.
+        context: String,
+    },
+    /// An outbound response about to be sent.
+    Outbound {
+        user_id: String,
+        channel: String,
+        content: String,
+        thread_id: Option<String>,
+    },
+    /// A new session was created.
+    SessionStart { user_id: String, session_id: String },
+    /// A session was ended (pruned).
+    SessionEnd { user_id: String, session_id: String },
+    /// The final response is being transformed before completing a turn.
+    ResponseTransform {
+        user_id: String,
+        thread_id: String,
+        response: String,
+    },
+}
+
+impl HookEvent {
+    /// Returns the [`HookPoint`] this event corresponds to.
+    pub fn hook_point(&self) -> HookPoint {
+        match self {
+            HookEvent::Inbound { .. } => HookPoint::BeforeInbound,
+            HookEvent::ToolCall { .. } => HookPoint::BeforeToolCall,
+            HookEvent::Outbound { .. } => HookPoint::BeforeOutbound,
+            HookEvent::SessionStart { .. } => HookPoint::OnSessionStart,
+            HookEvent::SessionEnd { .. } => HookPoint::OnSessionEnd,
+            HookEvent::ResponseTransform { .. } => HookPoint::TransformResponse,
+        }
+    }
+
+    /// Apply a modification string to the event's primary content field.
+    pub fn apply_modification(&mut self, modified: &str) {
+        match self {
+            HookEvent::Inbound { content, .. } | HookEvent::Outbound { content, .. } => {
+                *content = modified.to_string();
+            }
+            HookEvent::ToolCall { parameters, .. } => match serde_json::from_str(modified) {
+                Ok(parsed) => *parameters = parsed,
+                Err(e) => {
+                    tracing::warn!(
+                        "Hook returned non-JSON modification for ToolCall, ignoring: {}",
+                        e
+                    );
+                }
+            },
+            HookEvent::ResponseTransform { response, .. } => {
+                *response = modified.to_string();
+            }
+            HookEvent::SessionStart { .. } | HookEvent::SessionEnd { .. } => {
+                // Session events don't have modifiable content
+            }
+        }
+    }
+}
+
+/// The result of executing a hook.
+#[derive(Debug, Clone)]
+pub enum HookOutcome {
+    /// Continue processing, optionally with modified content.
+    Continue {
+        /// If `Some`, replace the event's primary content with this value.
+        modified: Option<String>,
+    },
+    /// Reject the event entirely.
+    Reject {
+        /// Human-readable reason for the rejection.
+        reason: String,
+    },
+}
+
+impl HookOutcome {
+    /// Shorthand for `Continue { modified: None }`.
+    pub fn ok() -> Self {
+        HookOutcome::Continue { modified: None }
+    }
+
+    /// Shorthand for `Continue { modified: Some(value) }`.
+    pub fn modify(value: String) -> Self {
+        HookOutcome::Continue {
+            modified: Some(value),
+        }
+    }
+
+    /// Shorthand for `Reject { reason }`.
+    pub fn reject(reason: impl Into<String>) -> Self {
+        HookOutcome::Reject {
+            reason: reason.into(),
+        }
+    }
+}
+
+/// How to handle hook execution failures.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum HookFailureMode {
+    /// On error/timeout, continue processing as if the hook returned `ok()`.
+    FailOpen,
+    /// On error/timeout, reject the event.
+    FailClosed,
+}
+
+/// Hook execution errors.
+#[derive(Debug, thiserror::Error)]
+pub enum HookError {
+    #[error("Hook execution failed: {reason}")]
+    ExecutionFailed { reason: String },
+
+    #[error("Hook timed out after {timeout:?}")]
+    Timeout { timeout: Duration },
+
+    #[error("Hook rejected: {reason}")]
+    Rejected { reason: String },
+}
+
+/// Context passed to hooks alongside the event.
+pub struct HookContext {
+    /// Arbitrary metadata hooks can use.
+    pub metadata: serde_json::Value,
+}
+
+impl Default for HookContext {
+    fn default() -> Self {
+        Self {
+            metadata: serde_json::Value::Null,
+        }
+    }
+}
+
+/// Trait for implementing lifecycle hooks.
+///
+/// Hooks intercept and can modify agent operations at well-defined points.
+#[async_trait]
+pub trait Hook: Send + Sync {
+    /// A unique name for this hook.
+    fn name(&self) -> &str;
+
+    /// The lifecycle points this hook should be called at.
+    fn hook_points(&self) -> &[HookPoint];
+
+    /// How to handle failures in this hook.
+    ///
+    /// Default: `FailOpen` (continue on error).
+    fn failure_mode(&self) -> HookFailureMode {
+        HookFailureMode::FailOpen
+    }
+
+    /// Maximum time this hook is allowed to run.
+    ///
+    /// Default: 5 seconds.
+    fn timeout(&self) -> Duration {
+        Duration::from_secs(5)
+    }
+
+    /// Execute the hook.
+    async fn execute(&self, event: &HookEvent, ctx: &HookContext)
+    -> Result<HookOutcome, HookError>;
+}

src/hooks/mod.rs

@@ -0,0 +1,19 @@
+//! Lifecycle hooks for intercepting and transforming agent operations.
+//!
+//! The hook system provides 6 well-defined interception points:
+//!
+//! - **BeforeInbound** — Before processing an inbound user message
+//! - **BeforeToolCall** — Before executing a tool call
+//! - **BeforeOutbound** — Before sending an outbound response
+//! - **OnSessionStart** — When a new session starts
+//! - **OnSessionEnd** — When a session ends
+//! - **TransformResponse** — Transform the final response before completing a turn
+//!
+//! Hooks are executed in priority order (lower number = higher priority).
+//! Each hook can pass through, modify content, or reject the event.
+
+pub mod hook;
+pub mod registry;
+
+pub use hook::{Hook, HookContext, HookError, HookEvent, HookFailureMode, HookOutcome, HookPoint};
+pub use registry::HookRegistry;

src/hooks/registry.rs

@@ -0,0 +1,555 @@
+//! Hook registry for managing and executing lifecycle hooks.
+
+use std::sync::Arc;
+
+use tokio::sync::RwLock;
+
+use crate::hooks::hook::{Hook, HookContext, HookError, HookEvent, HookFailureMode, HookOutcome};
+
+/// A registered hook with its priority.
+struct HookEntry {
+    hook: Arc<dyn Hook>,
+    priority: u32,
+}
+
+/// Registry that manages hooks and executes them at lifecycle points.
+///
+/// Hooks are executed in priority order (lower number = higher priority).
+/// A `Reject` outcome stops the chain immediately.
+/// A `Modify` outcome chains through subsequent hooks.
+pub struct HookRegistry {
+    hooks: RwLock<Vec<HookEntry>>,
+}
+
+impl HookRegistry {
+    /// Create an empty registry.
+    pub fn new() -> Self {
+        Self {
+            hooks: RwLock::new(Vec::new()),
+        }
+    }
+
+    /// Register a hook with default priority (100).
+    pub async fn register(&self, hook: Arc<dyn Hook>) {
+        self.register_with_priority(hook, 100).await;
+    }
+
+    /// Register a hook with a specific priority.
+    ///
+    /// Lower priority number = runs first.
+    pub async fn register_with_priority(&self, hook: Arc<dyn Hook>, priority: u32) {
+        let mut hooks = self.hooks.write().await;
+        hooks.push(HookEntry { hook, priority });
+        hooks.sort_by_key(|e| e.priority);
+    }
+
+    /// Unregister a hook by name. Returns `true` if it was found and removed.
+    pub async fn unregister(&self, name: &str) -> bool {
+        let mut hooks = self.hooks.write().await;
+        let before = hooks.len();
+        hooks.retain(|e| e.hook.name() != name);
+        hooks.len() < before
+    }
+
+    /// List all registered hook names (in priority order).
+    pub async fn list(&self) -> Vec<String> {
+        let hooks = self.hooks.read().await;
+        hooks.iter().map(|e| e.hook.name().to_string()).collect()
+    }
+
+    /// Run all hooks matching the event's hook point.
+    ///
+    /// - Hooks run in priority order (lowest first).
+    /// - `Reject` stops the chain immediately.
+    /// - `Modify` chains the modification through subsequent hooks.
+    /// - Timeout/error handling respects each hook's `failure_mode`.
+    pub async fn run(&self, event: &HookEvent) -> Result<HookOutcome, HookError> {
+        let point = event.hook_point();
+        let ctx = HookContext::default();
+
+        // Clone matching hooks and drop the read guard before executing.
+        // Each hook can run up to its timeout, so holding the guard would
+        // block concurrent register/unregister/run calls.
+        let matching: Vec<Arc<dyn Hook>> = {
+            let hooks = self.hooks.read().await;
+            hooks
+                .iter()
+                .filter(|e| e.hook.hook_points().contains(&point))
+                .map(|e| e.hook.clone())
+                .collect()
+        };
+
+        if matching.is_empty() {
+            return Ok(HookOutcome::ok());
+        }
+
+        let mut current_event = event.clone();
+
+        for hook in &matching {
+            let timeout = hook.timeout();
+
+            let result = tokio::time::timeout(timeout, hook.execute(&current_event, &ctx)).await;
+
+            match result {
+                Ok(Ok(HookOutcome::Reject { reason })) => {
+                    tracing::debug!(hook = hook.name(), "Hook rejected: {}", reason);
+                    return Err(HookError::Rejected { reason });
+                }
+                Ok(Ok(HookOutcome::Continue {
+                    modified: Some(value),
+                })) => {
+                    tracing::debug!(hook = hook.name(), "Hook modified content");
+                    current_event.apply_modification(&value);
+                }
+                Ok(Ok(HookOutcome::Continue { modified: None })) => {
+                    // No-op, continue chain
+                }
+                Ok(Err(err)) => match hook.failure_mode() {
+                    HookFailureMode::FailOpen => {
+                        tracing::warn!(hook = hook.name(), "Hook failed (fail-open): {}", err);
+                    }
+                    HookFailureMode::FailClosed => {
+                        tracing::warn!(hook = hook.name(), "Hook failed (fail-closed): {}", err);
+                        return Err(HookError::ExecutionFailed {
+                            reason: format!("Hook '{}' failed: {}", hook.name(), err),
+                        });
+                    }
+                },
+                Err(_elapsed) => match hook.failure_mode() {
+                    HookFailureMode::FailOpen => {
+                        tracing::warn!(
+                            hook = hook.name(),
+                            "Hook timed out (fail-open) after {:?}",
+                            timeout
+                        );
+                    }
+                    HookFailureMode::FailClosed => {
+                        tracing::warn!(
+                            hook = hook.name(),
+                            "Hook timed out (fail-closed) after {:?}",
+                            timeout
+                        );
+                        return Err(HookError::Timeout { timeout });
+                    }
+                },
+            }
+        }
+
+        // Determine final outcome by comparing with original event
+        let modified = extract_content(&current_event);
+        let original = extract_content(event);
+
+        if modified != original {
+            Ok(HookOutcome::modify(modified))
+        } else {
+            Ok(HookOutcome::ok())
+        }
+    }
+}
+
+impl Default for HookRegistry {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+/// Extract the primary content string from a hook event.
+fn extract_content(event: &HookEvent) -> String {
+    match event {
+        HookEvent::Inbound { content, .. } | HookEvent::Outbound { content, .. } => content.clone(),
+        HookEvent::ToolCall { parameters, .. } => {
+            serde_json::to_string(parameters).unwrap_or_default()
+        }
+        HookEvent::ResponseTransform { response, .. } => response.clone(),
+        HookEvent::SessionStart { session_id, .. } | HookEvent::SessionEnd { session_id, .. } => {
+            session_id.clone()
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::hooks::hook::{HookFailureMode, HookPoint};
+    use async_trait::async_trait;
+    use std::time::Duration;
+
+    /// A test hook that always returns ok.
+    struct PassthroughHook {
+        name: String,
+        points: Vec<HookPoint>,
+    }
+
+    #[async_trait]
+    impl Hook for PassthroughHook {
+        fn name(&self) -> &str {
+            &self.name
+        }
+        fn hook_points(&self) -> &[HookPoint] {
+            &self.points
+        }
+        async fn execute(
+            &self,
+            _event: &HookEvent,
+            _ctx: &HookContext,
+        ) -> Result<HookOutcome, HookError> {
+            Ok(HookOutcome::ok())
+        }
+    }
+
+    /// A hook that modifies content by appending a suffix.
+    struct ModifyHook {
+        name: String,
+        suffix: String,
+        points: Vec<HookPoint>,
+    }
+
+    #[async_trait]
+    impl Hook for ModifyHook {
+        fn name(&self) -> &str {
+            &self.name
+        }
+        fn hook_points(&self) -> &[HookPoint] {
+            &self.points
+        }
+        async fn execute(
+            &self,
+            event: &HookEvent,
+            _ctx: &HookContext,
+        ) -> Result<HookOutcome, HookError> {
+            let content = extract_content(event);
+            Ok(HookOutcome::modify(format!("{}{}", content, self.suffix)))
+        }
+    }
+
+    /// A hook that always rejects.
+    struct RejectHook {
+        name: String,
+        reason: String,
+        points: Vec<HookPoint>,
+    }
+
+    #[async_trait]
+    impl Hook for RejectHook {
+        fn name(&self) -> &str {
+            &self.name
+        }
+        fn hook_points(&self) -> &[HookPoint] {
+            &self.points
+        }
+        async fn execute(
+            &self,
+            _event: &HookEvent,
+            _ctx: &HookContext,
+        ) -> Result<HookOutcome, HookError> {
+            Ok(HookOutcome::reject(&self.reason))
+        }
+    }
+
+    /// A hook that always errors.
+    struct ErrorHook {
+        name: String,
+        points: Vec<HookPoint>,
+        failure_mode: HookFailureMode,
+    }
+
+    #[async_trait]
+    impl Hook for ErrorHook {
+        fn name(&self) -> &str {
+            &self.name
+        }
+        fn hook_points(&self) -> &[HookPoint] {
+            &self.points
+        }
+        fn failure_mode(&self) -> HookFailureMode {
+            self.failure_mode
+        }
+        async fn execute(
+            &self,
+            _event: &HookEvent,
+            _ctx: &HookContext,
+        ) -> Result<HookOutcome, HookError> {
+            Err(HookError::ExecutionFailed {
+                reason: "test error".into(),
+            })
+        }
+    }
+
+    /// A hook that sleeps longer than its timeout.
+    struct SlowHook {
+        name: String,
+        points: Vec<HookPoint>,
+        failure_mode: HookFailureMode,
+    }
+
+    #[async_trait]
+    impl Hook for SlowHook {
+        fn name(&self) -> &str {
+            &self.name
+        }
+        fn hook_points(&self) -> &[HookPoint] {
+            &self.points
+        }
+        fn failure_mode(&self) -> HookFailureMode {
+            self.failure_mode
+        }
+        fn timeout(&self) -> Duration {
+            Duration::from_millis(50)
+        }
+        async fn execute(
+            &self,
+            _event: &HookEvent,
+            _ctx: &HookContext,
+        ) -> Result<HookOutcome, HookError> {
+            tokio::time::sleep(Duration::from_millis(200)).await;
+            Ok(HookOutcome::ok())
+        }
+    }
+
+    fn test_event() -> HookEvent {
+        HookEvent::Inbound {
+            user_id: "user-1".into(),
+            channel: "test".into(),
+            content: "hello".into(),
+            thread_id: None,
+        }
+    }
+
+    #[tokio::test]
+    async fn test_empty_registry_returns_ok() {
+        let registry = HookRegistry::new();
+        let result = registry.run(&test_event()).await;
+        assert!(result.is_ok());
+        assert!(matches!(
+            result.unwrap(),
+            HookOutcome::Continue { modified: None }
+        ));
+    }
+
+    #[tokio::test]
+    async fn test_register_and_list() {
+        let registry = HookRegistry::new();
+        registry
+            .register(Arc::new(PassthroughHook {
+                name: "hook-a".into(),
+                points: vec![HookPoint::BeforeInbound],
+            }))
+            .await;
+        registry
+            .register(Arc::new(PassthroughHook {
+                name: "hook-b".into(),
+                points: vec![HookPoint::BeforeInbound],
+            }))
+            .await;
+
+        let names = registry.list().await;
+        assert_eq!(names, vec!["hook-a", "hook-b"]);
+    }
+
+    #[tokio::test]
+    async fn test_priority_ordering() {
+        let registry = HookRegistry::new();
+
+        // Register in reverse priority order
+        registry
+            .register_with_priority(
+                Arc::new(ModifyHook {
+                    name: "low-prio".into(),
+                    suffix: "-LOW".into(),
+                    points: vec![HookPoint::BeforeInbound],
+                }),
+                200,
+            )
+            .await;
+        registry
+            .register_with_priority(
+                Arc::new(ModifyHook {
+                    name: "high-prio".into(),
+                    suffix: "-HIGH".into(),
+                    points: vec![HookPoint::BeforeInbound],
+                }),
+                10,
+            )
+            .await;
+
+        // Should run in priority order: high-prio first, then low-prio
+        let names = registry.list().await;
+        assert_eq!(names[0], "high-prio");
+        assert_eq!(names[1], "low-prio");
+
+        let result = registry.run(&test_event()).await.unwrap();
+        match result {
+            HookOutcome::Continue { modified: Some(m) } => {
+                // "hello" -> "hello-HIGH" -> "hello-HIGH-LOW"
+                assert_eq!(m, "hello-HIGH-LOW");
+            }
+            other => panic!("Expected modification chain, got: {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn test_reject_stops_chain() {
+        let registry = HookRegistry::new();
+
+        registry
+            .register_with_priority(
+                Arc::new(RejectHook {
+                    name: "blocker".into(),
+                    reason: "blocked".into(),
+                    points: vec![HookPoint::BeforeInbound],
+                }),
+                10,
+            )
+            .await;
+        registry
+            .register_with_priority(
+                Arc::new(ModifyHook {
+                    name: "modifier".into(),
+                    suffix: "-MODIFIED".into(),
+                    points: vec![HookPoint::BeforeInbound],
+                }),
+                20,
+            )
+            .await;
+
+        let result = registry.run(&test_event()).await;
+        assert!(result.is_err());
+        match result.unwrap_err() {
+            HookError::Rejected { reason } => assert_eq!(reason, "blocked"),
+            other => panic!("Expected Rejected, got: {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn test_modification_chaining() {
+        let registry = HookRegistry::new();
+
+        registry
+            .register_with_priority(
+                Arc::new(ModifyHook {
+                    name: "first".into(),
+                    suffix: "-A".into(),
+                    points: vec![HookPoint::BeforeInbound],
+                }),
+                10,
+            )
+            .await;
+        registry
+            .register_with_priority(
+                Arc::new(ModifyHook {
+                    name: "second".into(),
+                    suffix: "-B".into(),
+                    points: vec![HookPoint::BeforeInbound],
+                }),
+                20,
+            )
+            .await;
+
+        let result = registry.run(&test_event()).await.unwrap();
+        match result {
+            HookOutcome::Continue { modified: Some(m) } => {
+                assert_eq!(m, "hello-A-B");
+            }
+            other => panic!("Expected chained modification, got: {:?}", other),
+        }
+    }
+
+    #[tokio::test]
+    async fn test_fail_open_on_error() {
+        let registry = HookRegistry::new();
+        registry
+            .register(Arc::new(ErrorHook {
+                name: "err-open".into(),
+                points: vec![HookPoint::BeforeInbound],
+                failure_mode: HookFailureMode::FailOpen,
+            }))
+            .await;
+
+        let result = registry.run(&test_event()).await;
+        assert!(result.is_ok());
+    }
+
+    #[tokio::test]
+    async fn test_fail_closed_on_error() {
+        let registry = HookRegistry::new();
+        registry
+            .register(Arc::new(ErrorHook {
+                name: "err-closed".into(),
+                points: vec![HookPoint::BeforeInbound],
+                failure_mode: HookFailureMode::FailClosed,
+            }))
+            .await;
+
+        let result = registry.run(&test_event()).await;
+        assert!(result.is_err());
+        assert!(matches!(
+            result.unwrap_err(),
+            HookError::ExecutionFailed { .. }
+        ));
+    }
+
+    #[tokio::test]
+    async fn test_fail_open_on_timeout() {
+        let registry = HookRegistry::new();
+        registry
+            .register(Arc::new(SlowHook {
+                name: "slow-open".into(),
+                points: vec![HookPoint::BeforeInbound],
+                failure_mode: HookFailureMode::FailOpen,
+            }))
+            .await;
+
+        let result = registry.run(&test_event()).await;
+        assert!(result.is_ok());
+    }
+
+    #[tokio::test]
+    async fn test_fail_closed_on_timeout() {
+        let registry = HookRegistry::new();
+        registry
+            .register(Arc::new(SlowHook {
+                name: "slow-closed".into(),
+                points: vec![HookPoint::BeforeInbound],
+                failure_mode: HookFailureMode::FailClosed,
+            }))
+            .await;
+
+        let result = registry.run(&test_event()).await;
+        assert!(result.is_err());
+        assert!(matches!(result.unwrap_err(), HookError::Timeout { .. }));
+    }
+
+    #[tokio::test]
+    async fn test_unregister() {
+        let registry = HookRegistry::new();
+        registry
+            .register(Arc::new(PassthroughHook {
+                name: "removable".into(),
+                points: vec![HookPoint::BeforeInbound],
+            }))
+            .await;
+
+        assert_eq!(registry.list().await.len(), 1);
+        assert!(registry.unregister("removable").await);
+        assert_eq!(registry.list().await.len(), 0);
+
+        // Unregistering non-existent returns false
+        assert!(!registry.unregister("nonexistent").await);
+    }
+
+    #[tokio::test]
+    async fn test_hooks_only_match_their_points() {
+        let registry = HookRegistry::new();
+        registry
+            .register(Arc::new(RejectHook {
+                name: "outbound-only".into(),
+                reason: "blocked".into(),
+                points: vec![HookPoint::BeforeOutbound],
+            }))
+            .await;
+
+        // Inbound event should not be affected by outbound-only hook
+        let result = registry.run(&test_event()).await;
+        assert!(result.is_ok());
+    }
+}

src/lib.rs

@@ -39,16 +39,19 @@
 //! - **Continuous learning** - Improve estimates from historical data
 
 pub mod agent;
+pub mod boot_screen;
 pub mod bootstrap;
 pub mod channels;
 pub mod cli;
 pub mod config;
 pub mod context;
+pub mod db;
 pub mod error;
 pub mod estimation;
 pub mod evaluation;
 pub mod extensions;
 pub mod history;
+pub mod hooks;
 pub mod llm;
 pub mod orchestrator;
 pub mod pairing;
@@ -58,6 +61,8 @@ pub mod secrets;
 pub mod settings;
 pub mod setup;
 pub mod tools;
+pub mod tracing_fmt;
+pub mod util;
 pub mod worker;
 pub mod workspace;
 

src/llm/mod.rs

@@ -8,20 +8,26 @@
 //! - **OpenAI-compatible**: Any endpoint that speaks the OpenAI API
 
 mod costs;
+pub mod failover;
 mod nearai;
 mod nearai_chat;
 mod provider;
 mod reasoning;
+mod retry;
 mod rig_adapter;
 pub mod session;
 
+pub use failover::{CooldownConfig, FailoverProvider};
 pub use nearai::{ModelInfo, NearAiProvider};
 pub use nearai_chat::NearAiChatProvider;
 pub use provider::{
     ChatMessage, CompletionRequest, CompletionResponse, FinishReason, LlmProvider, ModelMetadata,
     Role, ToolCall, ToolCompletionRequest, ToolCompletionResponse, ToolDefinition, ToolResult,
 };
-pub use reasoning::{ActionPlan, Reasoning, ReasoningContext, RespondResult, ToolSelection};
+pub use reasoning::{
+    ActionPlan, Reasoning, ReasoningContext, RespondOutput, RespondResult, TokenUsage,
+    ToolSelection,
+};
 pub use rig_adapter::RigAdapter;
 pub use session::{SessionConfig, SessionManager, create_session_manager};
 
@@ -30,7 +36,7 @@ use std::sync::Arc;
 use rig::client::CompletionClient;
 use secrecy::ExposeSecret;
 
-use crate::config::{LlmBackend, LlmConfig, NearAiApiMode};
+use crate::config::{LlmBackend, LlmConfig, NearAiApiMode, NearAiConfig};
 use crate::error::LlmError;
 
 /// Create an LLM provider based on configuration.
@@ -43,7 +49,7 @@ pub fn create_llm_provider(
     session: Arc<SessionManager>,
 ) -> Result<Arc<dyn LlmProvider>, LlmError> {
     match config.backend {
-        LlmBackend::NearAi => create_nearai_provider(config, session),
+        LlmBackend::NearAi => create_llm_provider_with_config(&config.nearai, session),
         LlmBackend::OpenAi => create_openai_provider(config),
         LlmBackend::Anthropic => create_anthropic_provider(config),
         LlmBackend::Ollama => create_ollama_provider(config),
@@ -51,21 +57,28 @@ pub fn create_llm_provider(
     }
 }
 
-fn create_nearai_provider(
-    config: &LlmConfig,
+/// Create an LLM provider from a `NearAiConfig` directly.
+///
+/// This is useful when constructing additional providers for failover,
+/// where only the model name differs from the primary config.
+pub fn create_llm_provider_with_config(
+    config: &NearAiConfig,
     session: Arc<SessionManager>,
 ) -> Result<Arc<dyn LlmProvider>, LlmError> {
-    match config.nearai.api_mode {
+    match config.api_mode {
         NearAiApiMode::Responses => {
-            tracing::info!("Using NEAR AI Responses API (chat-api) with session auth");
-            Ok(Arc::new(NearAiProvider::new(
-                config.nearai.clone(),
-                session,
-            )))
+            tracing::info!(
+                model = %config.model,
+                "Using Responses API (chat-api) with session auth"
+            );
+            Ok(Arc::new(NearAiProvider::new(config.clone(), session)))
         }
         NearAiApiMode::ChatCompletions => {
-            tracing::info!("Using NEAR AI Chat Completions API (cloud-api) with API key auth");
-            Ok(Arc::new(NearAiChatProvider::new(config.nearai.clone())?))
+            tracing::info!(
+                model = %config.model,
+                "Using Chat Completions API (cloud-api) with API key auth"
+            );
+            Ok(Arc::new(NearAiChatProvider::new(config.clone())?))
         }
     }
 }
@@ -170,3 +183,108 @@ fn create_openai_compatible_provider(config: &LlmConfig) -> Result<Arc<dyn LlmPr
     );
     Ok(Arc::new(RigAdapter::new(model, &compat.model)))
 }
+
+/// Create a cheap/fast LLM provider for lightweight tasks (heartbeat, routing, evaluation).
+///
+/// Uses `NEARAI_CHEAP_MODEL` if set, otherwise falls back to the main provider.
+/// Currently only supports NEAR AI backends (Responses and ChatCompletions modes).
+pub fn create_cheap_llm_provider(
+    config: &LlmConfig,
+    session: Arc<SessionManager>,
+) -> Result<Option<Arc<dyn LlmProvider>>, LlmError> {
+    let Some(ref cheap_model) = config.nearai.cheap_model else {
+        return Ok(None);
+    };
+
+    if config.backend != LlmBackend::NearAi {
+        tracing::warn!(
+            "NEARAI_CHEAP_MODEL is set but LLM_BACKEND is {:?}, not NearAi. \
+             Cheap model setting will be ignored.",
+            config.backend
+        );
+        return Ok(None);
+    }
+
+    let mut cheap_config = config.nearai.clone();
+    cheap_config.model = cheap_model.clone();
+
+    tracing::info!("Cheap LLM provider: {}", cheap_model);
+
+    match cheap_config.api_mode {
+        NearAiApiMode::Responses => Ok(Some(Arc::new(NearAiProvider::new(cheap_config, session)))),
+        NearAiApiMode::ChatCompletions => {
+            Ok(Some(Arc::new(NearAiChatProvider::new(cheap_config)?)))
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::config::{LlmBackend, NearAiApiMode, NearAiConfig};
+    use std::path::PathBuf;
+
+    fn test_nearai_config() -> NearAiConfig {
+        NearAiConfig {
+            model: "test-model".to_string(),
+            cheap_model: None,
+            base_url: "https://api.near.ai".to_string(),
+            auth_base_url: "https://private.near.ai".to_string(),
+            session_path: PathBuf::from("/tmp/test-session.json"),
+            api_mode: NearAiApiMode::Responses,
+            api_key: None,
+            fallback_model: None,
+            max_retries: 3,
+            failover_cooldown_secs: 300,
+            failover_cooldown_threshold: 3,
+        }
+    }
+
+    fn test_llm_config() -> LlmConfig {
+        LlmConfig {
+            backend: LlmBackend::NearAi,
+            nearai: test_nearai_config(),
+            openai: None,
+            anthropic: None,
+            ollama: None,
+            openai_compatible: None,
+        }
+    }
+
+    #[test]
+    fn test_create_cheap_llm_provider_returns_none_when_not_configured() {
+        let config = test_llm_config();
+        let session = Arc::new(SessionManager::new(SessionConfig::default()));
+
+        let result = create_cheap_llm_provider(&config, session);
+        assert!(result.is_ok());
+        assert!(result.unwrap().is_none());
+    }
+
+    #[test]
+    fn test_create_cheap_llm_provider_creates_provider_when_configured() {
+        let mut config = test_llm_config();
+        config.nearai.cheap_model = Some("cheap-test-model".to_string());
+
+        let session = Arc::new(SessionManager::new(SessionConfig::default()));
+        let result = create_cheap_llm_provider(&config, session);
+
+        assert!(result.is_ok());
+        let provider = result.unwrap();
+        assert!(provider.is_some());
+        assert_eq!(provider.unwrap().model_name(), "cheap-test-model");
+    }
+
+    #[test]
+    fn test_create_cheap_llm_provider_ignored_for_non_nearai_backend() {
+        let mut config = test_llm_config();
+        config.backend = LlmBackend::OpenAi;
+        config.nearai.cheap_model = Some("cheap-test-model".to_string());
+
+        let session = Arc::new(SessionManager::new(SessionConfig::default()));
+        let result = create_cheap_llm_provider(&config, session);
+
+        assert!(result.is_ok());
+        assert!(result.unwrap().is_none());
+    }
+}

src/llm/nearai.rs

@@ -19,6 +19,7 @@ use crate::llm::provider::{
     ChatMessage, CompletionRequest, CompletionResponse, FinishReason, LlmProvider, Role, ToolCall,
     ToolCompletionRequest, ToolCompletionResponse,
 };
+use crate::llm::retry::{is_retryable_status, retry_backoff_delay};
 use crate::llm::session::SessionManager;
 
 /// Information about an available model from NEAR AI API.
@@ -209,20 +210,20 @@ impl NearAiProvider {
             data: Option<Vec<ModelEntry>>,
         }
 
-        if let Ok(resp) = serde_json::from_str::<ModelsResponse>(&response_text) {
-            if let Some(entries) = resp.models.or(resp.data) {
-                let models: Vec<ModelInfo> = entries
-                    .into_iter()
-                    .filter_map(|e| {
-                        e.get_name().map(|name| ModelInfo {
-                            name,
-                            provider: None,
-                        })
+        if let Ok(resp) = serde_json::from_str::<ModelsResponse>(&response_text)
+            && let Some(entries) = resp.models.or(resp.data)
+        {
+            let models: Vec<ModelInfo> = entries
+                .into_iter()
+                .filter_map(|e| {
+                    e.get_name().map(|name| ModelInfo {
+                        name,
+                        provider: None,
                     })
-                    .collect();
-                if !models.is_empty() {
-                    return Ok(models);
-                }
+                })
+                .collect();
+            if !models.is_empty() {
+                return Ok(models);
             }
         }
 
@@ -270,88 +271,139 @@ impl NearAiProvider {
         }
     }
 
-    /// Inner request implementation without retry logic.
+    /// Inner request implementation with retry logic for transient errors.
+    ///
+    /// Retries on HTTP 429, 500, 502, 503, 504 with exponential backoff.
+    /// Does not retry on client errors (400, 401, 403, 404) or parse errors.
     async fn send_request_inner<T: Serialize + std::fmt::Debug, R: for<'de> Deserialize<'de>>(
         &self,
         path: &str,
         body: &T,
     ) -> Result<R, LlmError> {
         let url = self.api_url(path);
-        let token = self.session.get_token().await?;
+        let max_retries = self.config.max_retries;
 
-        tracing::debug!("Sending request to NEAR AI: {}", url);
-        tracing::debug!("Request body: {:?}", body);
+        for attempt in 0..=max_retries {
+            let token = self.session.get_token().await?;
 
-        let response = self
-            .client
-            .post(&url)
-            .header("Authorization", format!("Bearer {}", token.expose_secret()))
-            .header("Content-Type", "application/json")
-            .json(body)
-            .send()
-            .await
-            .map_err(|e| {
-                tracing::error!("NEAR AI request failed: {}", e);
-                e
-            })?;
+            tracing::debug!(
+                "Sending request to NEAR AI: {} (attempt {})",
+                url,
+                attempt + 1
+            );
+            tracing::debug!("Request body: {:?}", body);
 
-        let status = response.status();
-        let response_text = response.text().await.unwrap_or_default();
+            let response = self
+                .client
+                .post(&url)
+                .header("Authorization", format!("Bearer {}", token.expose_secret()))
+                .header("Content-Type", "application/json")
+                .json(body)
+                .send()
+                .await;
 
-        tracing::debug!("NEAR AI response status: {}", status);
-        tracing::debug!("NEAR AI response body: {}", response_text);
+            let response = match response {
+                Ok(r) => r,
+                Err(e) => {
+                    tracing::error!("NEAR AI request failed: {}", e);
+                    // Network errors (timeout, connection refused) are transient
+                    if attempt < max_retries {
+                        let delay = retry_backoff_delay(attempt);
+                        tracing::warn!(
+                            "NEAR AI request error (attempt {}/{}), retrying in {:?}: {}",
+                            attempt + 1,
+                            max_retries + 1,
+                            delay,
+                            e,
+                        );
+                        tokio::time::sleep(delay).await;
+                        continue;
+                    }
+                    return Err(e.into());
+                }
+            };
 
-        if !status.is_success() {
-            // Check for session expiration (401 with specific message patterns)
-            if status.as_u16() == 401 {
-                let is_session_expired = response_text.to_lowercase().contains("session")
-                    && (response_text.to_lowercase().contains("expired")
-                        || response_text.to_lowercase().contains("invalid"));
+            let status = response.status();
+            let response_text = response.text().await.unwrap_or_default();
 
-                if is_session_expired {
-                    return Err(LlmError::SessionExpired {
+            tracing::debug!("NEAR AI response status: {}", status);
+            tracing::debug!("NEAR AI response body: {}", response_text);
+
+            if !status.is_success() {
+                let status_code = status.as_u16();
+
+                // Check for session expiration (401 with specific message patterns)
+                if status_code == 401 {
+                    let lower = response_text.to_lowercase();
+                    let is_session_expired = lower.contains("session")
+                        && (lower.contains("expired") || lower.contains("invalid"));
+
+                    if is_session_expired {
+                        return Err(LlmError::SessionExpired {
+                            provider: "nearai".to_string(),
+                        });
+                    }
+
+                    // Generic 401 -- not retryable
+                    return Err(LlmError::AuthFailed {
                         provider: "nearai".to_string(),
                     });
                 }
 
-                // Generic 401 without session expiration indication
-                return Err(LlmError::AuthFailed {
-                    provider: "nearai".to_string(),
-                });
-            }
+                // Check if this is a transient error worth retrying
+                if is_retryable_status(status_code) && attempt < max_retries {
+                    let delay = retry_backoff_delay(attempt);
+                    tracing::warn!(
+                        "NEAR AI returned HTTP {} (attempt {}/{}), retrying in {:?}",
+                        status_code,
+                        attempt + 1,
+                        max_retries + 1,
+                        delay,
+                    );
+                    tokio::time::sleep(delay).await;
+                    continue;
+                }
 
-            // Try to parse as JSON error
-            if let Ok(error) = serde_json::from_str::<NearAiErrorResponse>(&response_text) {
-                if status.as_u16() == 429 {
-                    return Err(LlmError::RateLimited {
+                // Non-retryable error or exhausted retries
+                if let Ok(error) = serde_json::from_str::<NearAiErrorResponse>(&response_text) {
+                    if status_code == 429 {
+                        return Err(LlmError::RateLimited {
+                            provider: "nearai".to_string(),
+                            retry_after: None,
+                        });
+                    }
+                    return Err(LlmError::RequestFailed {
                         provider: "nearai".to_string(),
-                        retry_after: None,
+                        reason: error.error,
                     });
                 }
+
                 return Err(LlmError::RequestFailed {
                     provider: "nearai".to_string(),
-                    reason: error.error,
+                    reason: format!("HTTP {}: {}", status, response_text),
                 });
             }
 
-            return Err(LlmError::RequestFailed {
-                provider: "nearai".to_string(),
-                reason: format!("HTTP {}: {}", status, response_text),
-            });
+            // Success -- parse the response
+            return match serde_json::from_str::<R>(&response_text) {
+                Ok(parsed) => Ok(parsed),
+                Err(e) => {
+                    tracing::debug!("Response is not expected JSON format: {}", e);
+                    tracing::debug!("Will try alternative parsing in caller");
+                    Err(LlmError::InvalidResponse {
+                        provider: "nearai".to_string(),
+                        reason: format!("Parse error: {}. Raw: {}", e, response_text),
+                    })
+                }
+            };
         }
 
-        // Try to parse as our expected type
-        match serde_json::from_str::<R>(&response_text) {
-            Ok(parsed) => Ok(parsed),
-            Err(e) => {
-                tracing::debug!("Response is not expected JSON format: {}", e);
-                tracing::debug!("Will try alternative parsing in caller");
-                Err(LlmError::InvalidResponse {
-                    provider: "nearai".to_string(),
-                    reason: format!("Parse error: {}. Raw: {}", e, response_text),
-                })
-            }
-        }
+        // This is unreachable because the loop always returns, but the compiler
+        // cannot prove that. Return a generic error as a safety net.
+        Err(LlmError::RequestFailed {
+            provider: "nearai".to_string(),
+            reason: "retry loop exited unexpectedly".to_string(),
+        })
     }
 }
 
@@ -456,7 +508,7 @@ impl LlmProvider for NearAiProvider {
             Err(e) => return Err(e),
         };
 
-        tracing::debug!("NEAR AI response: {:?}", response);
+        tracing::debug!("NEAR AI response: output_items={}", response.output.len());
 
         // Extract text from response output
         // Try multiple formats since API response shape may vary
@@ -464,11 +516,6 @@ impl LlmProvider for NearAiProvider {
             .output
             .iter()
             .filter_map(|item| {
-                tracing::debug!(
-                    "Processing output item: type={}, text={:?}",
-                    item.item_type,
-                    item.text
-                );
                 if item.item_type == "message" {
                     // First check for direct text field on item
                     if let Some(ref text) = item.text {
@@ -479,11 +526,6 @@ impl LlmProvider for NearAiProvider {
                         contents
                             .iter()
                             .filter_map(|c| {
-                                tracing::debug!(
-                                    "Content item: type={}, text={:?}",
-                                    c.content_type,
-                                    c.text
-                                );
                                 // Accept various content types that might contain text
                                 match c.content_type.as_str() {
                                     "output_text" | "text" => c.text.clone(),
@@ -694,21 +736,21 @@ impl LlmProvider for NearAiProvider {
                         }
                     }
                 }
-            } else if item.item_type == "function_call" {
-                if let (Some(name), Some(call_id)) = (&item.name, &item.call_id) {
-                    // Parse arguments JSON string into Value
-                    let arguments = item
-                        .arguments
-                        .as_ref()
-                        .and_then(|s| serde_json::from_str(s).ok())
-                        .unwrap_or(serde_json::Value::Object(Default::default()));
+            } else if item.item_type == "function_call"
+                && let (Some(name), Some(call_id)) = (&item.name, &item.call_id)
+            {
+                // Parse arguments JSON string into Value
+                let arguments = item
+                    .arguments
+                    .as_ref()
+                    .and_then(|s| serde_json::from_str(s).ok())
+                    .unwrap_or(serde_json::Value::Object(Default::default()));
 
-                    tool_calls.push(ToolCall {
-                        id: call_id.clone(),
-                        name: name.clone(),
-                        arguments,
-                    });
-                }
+                tool_calls.push(ToolCall {
+                    id: call_id.clone(),
+                    name: name.clone(),
+                    arguments,
+                });
             }
         }
 

src/llm/nearai_chat.rs

@@ -16,6 +16,7 @@ use crate::llm::provider::{
     ChatMessage, CompletionRequest, CompletionResponse, FinishReason, LlmProvider, ModelMetadata,
     Role, ToolCall, ToolCompletionRequest, ToolCompletionResponse,
 };
+use crate::llm::retry::{is_retryable_status, retry_backoff_delay};
 
 /// NEAR AI Chat Completions API provider.
 pub struct NearAiChatProvider {
@@ -62,63 +63,116 @@ impl NearAiChatProvider {
             .unwrap_or_default()
     }
 
-    /// Send a request to the chat completions API.
+    /// Send a request to the chat completions API with retry on transient errors.
+    ///
+    /// Retries on HTTP 429, 500, 502, 503, 504 with exponential backoff.
+    /// Does not retry on client errors (400, 401, 403, 404) or parse errors.
     async fn send_request<T: Serialize, R: for<'de> Deserialize<'de>>(
         &self,
         body: &T,
     ) -> Result<R, LlmError> {
         let url = self.api_url("chat/completions");
+        let max_retries = self.config.max_retries;
 
-        tracing::debug!("Sending request to NEAR AI Chat: {}", url);
+        for attempt in 0..=max_retries {
+            tracing::debug!(
+                "Sending request to NEAR AI Chat: {} (attempt {})",
+                url,
+                attempt + 1,
+            );
 
-        // Log the request body for debugging tool call issues
-        if let Ok(json) = serde_json::to_string(body) {
-            tracing::debug!("NEAR AI Chat request body: {}", json);
-        }
+            if tracing::enabled!(tracing::Level::DEBUG)
+                && let Ok(json) = serde_json::to_string(body)
+            {
+                tracing::debug!("NEAR AI Chat request body: {}", json);
+            }
 
-        let response = self
-            .client
-            .post(&url)
-            .header("Authorization", format!("Bearer {}", self.api_key()))
-            .header("Content-Type", "application/json")
-            .json(body)
-            .send()
-            .await
-            .map_err(|e| {
-                tracing::error!("NEAR AI Chat request failed: {}", e);
-                LlmError::RequestFailed {
-                    provider: "nearai_chat".to_string(),
-                    reason: e.to_string(),
+            let response = self
+                .client
+                .post(&url)
+                .header("Authorization", format!("Bearer {}", self.api_key()))
+                .header("Content-Type", "application/json")
+                .json(body)
+                .send()
+                .await;
+
+            let response = match response {
+                Ok(r) => r,
+                Err(e) => {
+                    tracing::error!("NEAR AI Chat request failed: {}", e);
+                    if attempt < max_retries {
+                        let delay = retry_backoff_delay(attempt);
+                        tracing::warn!(
+                            "NEAR AI Chat request error (attempt {}/{}), retrying in {:?}: {}",
+                            attempt + 1,
+                            max_retries + 1,
+                            delay,
+                            e,
+                        );
+                        tokio::time::sleep(delay).await;
+                        continue;
+                    }
+                    return Err(LlmError::RequestFailed {
+                        provider: "nearai_chat".to_string(),
+                        reason: e.to_string(),
+                    });
                 }
-            })?;
+            };
 
-        let status = response.status();
-        let response_text = response.text().await.unwrap_or_default();
+            let status = response.status();
+            let response_text = response.text().await.unwrap_or_default();
 
-        tracing::debug!("NEAR AI Chat response status: {}", status);
-        tracing::debug!("NEAR AI Chat response body: {}", response_text);
+            tracing::debug!("NEAR AI Chat response status: {}", status);
+            tracing::debug!("NEAR AI Chat response body: {}", response_text);
 
-        if !status.is_success() {
-            if status.as_u16() == 401 {
-                return Err(LlmError::AuthFailed {
+            if !status.is_success() {
+                let status_code = status.as_u16();
+
+                // Auth errors are not retryable
+                if status_code == 401 {
+                    return Err(LlmError::AuthFailed {
+                        provider: "nearai_chat".to_string(),
+                    });
+                }
+
+                // Transient errors: retry with backoff
+                if is_retryable_status(status_code) && attempt < max_retries {
+                    let delay = retry_backoff_delay(attempt);
+                    tracing::warn!(
+                        "NEAR AI Chat returned HTTP {} (attempt {}/{}), retrying in {:?}",
+                        status_code,
+                        attempt + 1,
+                        max_retries + 1,
+                        delay,
+                    );
+                    tokio::time::sleep(delay).await;
+                    continue;
+                }
+
+                // Non-retryable or exhausted retries
+                if status_code == 429 {
+                    return Err(LlmError::RateLimited {
+                        provider: "nearai_chat".to_string(),
+                        retry_after: None,
+                    });
+                }
+                return Err(LlmError::RequestFailed {
                     provider: "nearai_chat".to_string(),
+                    reason: format!("HTTP {}: {}", status, response_text),
                 });
             }
-            if status.as_u16() == 429 {
-                return Err(LlmError::RateLimited {
-                    provider: "nearai_chat".to_string(),
-                    retry_after: None,
-                });
-            }
-            return Err(LlmError::RequestFailed {
+
+            // Success — parse the response
+            return serde_json::from_str(&response_text).map_err(|e| LlmError::InvalidResponse {
                 provider: "nearai_chat".to_string(),
-                reason: format!("HTTP {}: {}", status, response_text),
+                reason: format!("JSON parse error: {}. Raw: {}", e, response_text),
             });
         }
 
-        serde_json::from_str(&response_text).map_err(|e| LlmError::InvalidResponse {
+        // Safety net: unreachable because the loop always returns
+        Err(LlmError::RequestFailed {
             provider: "nearai_chat".to_string(),
-            reason: format!("JSON parse error: {}. Raw: {}", e, response_text),
+            reason: "retry loop exited unexpectedly".to_string(),
         })
     }
 
@@ -222,6 +276,12 @@ impl LlmProvider for NearAiChatProvider {
         let messages: Vec<ChatCompletionMessage> =
             req.messages.into_iter().map(|m| m.into()).collect();
 
+        // NEAR AI cloud-api does not support multi-turn tool calling (rejects
+        // any request containing role:"tool" messages with HTTP 400). Rewrite
+        // tool-call / tool-result pairs into plain text so the conversation
+        // history is preserved without using unsupported message roles.
+        let messages = flatten_tool_messages(messages);
+
         let tools: Vec<ChatCompletionTool> = req
             .tools
             .into_iter()
@@ -367,6 +427,64 @@ struct ChatCompletionMessage {
     tool_calls: Option<Vec<ChatCompletionToolCall>>,
 }
 
+/// Rewrite tool-call / tool-result messages into plain assistant/user text.
+///
+/// NEAR AI cloud-api does not support the OpenAI multi-turn tool-calling
+/// protocol (`role: "tool"` messages). This function converts:
+///   - Assistant messages with `tool_calls` → assistant text describing the calls
+///   - Tool result messages (`role: "tool"`) → user messages with the result
+///
+/// Non-tool messages pass through unchanged.
+fn flatten_tool_messages(messages: Vec<ChatCompletionMessage>) -> Vec<ChatCompletionMessage> {
+    let has_tool_msgs = messages.iter().any(|m| m.role == "tool");
+    if !has_tool_msgs {
+        return messages;
+    }
+
+    tracing::debug!("Flattening tool messages for NEAR AI compatibility");
+
+    messages
+        .into_iter()
+        .map(|msg| {
+            if let (true, Some(calls)) = (msg.role == "assistant", &msg.tool_calls) {
+                // Convert assistant tool_calls into descriptive text
+                let mut parts: Vec<String> = Vec::new();
+                if let Some(ref text) = msg.content
+                    && !text.is_empty()
+                {
+                    parts.push(text.clone());
+                }
+                for tc in calls {
+                    parts.push(format!(
+                        "[Called tool `{}` with arguments: {}]",
+                        tc.function.name, tc.function.arguments
+                    ));
+                }
+                ChatCompletionMessage {
+                    role: "assistant".to_string(),
+                    content: Some(parts.join("\n")),
+                    tool_call_id: None,
+                    name: None,
+                    tool_calls: None,
+                }
+            } else if msg.role == "tool" {
+                // Convert tool result into a user message
+                let tool_name = msg.name.as_deref().unwrap_or("unknown");
+                let result = msg.content.as_deref().unwrap_or("");
+                ChatCompletionMessage {
+                    role: "user".to_string(),
+                    content: Some(format!("[Tool `{}` returned: {}]", tool_name, result)),
+                    tool_call_id: None,
+                    name: None,
+                    tool_calls: None,
+                }
+            } else {
+                msg
+            }
+        })
+        .collect()
+}
+
 impl From<ChatMessage> for ChatCompletionMessage {
     fn from(msg: ChatMessage) -> Self {
         let role = match msg.role {
@@ -544,4 +662,119 @@ mod tests {
             serde_json::from_str(&calls[0].function.arguments).expect("valid JSON string");
         assert_eq!(parsed["key"], "value");
     }
+
+    #[test]
+    fn test_flatten_no_tool_messages_passthrough() {
+        let messages = vec![
+            ChatCompletionMessage {
+                role: "system".to_string(),
+                content: Some("You are helpful.".to_string()),
+                tool_call_id: None,
+                name: None,
+                tool_calls: None,
+            },
+            ChatCompletionMessage {
+                role: "user".to_string(),
+                content: Some("Hello".to_string()),
+                tool_call_id: None,
+                name: None,
+                tool_calls: None,
+            },
+        ];
+        let result = flatten_tool_messages(messages);
+        assert_eq!(result.len(), 2);
+        assert_eq!(result[0].role, "system");
+        assert_eq!(result[1].role, "user");
+    }
+
+    #[test]
+    fn test_flatten_tool_call_and_result() {
+        let messages = vec![
+            ChatCompletionMessage {
+                role: "user".to_string(),
+                content: Some("test".to_string()),
+                tool_call_id: None,
+                name: None,
+                tool_calls: None,
+            },
+            ChatCompletionMessage {
+                role: "assistant".to_string(),
+                content: None,
+                tool_call_id: None,
+                name: None,
+                tool_calls: Some(vec![ChatCompletionToolCall {
+                    id: "call_1".to_string(),
+                    call_type: "function".to_string(),
+                    function: ChatCompletionToolCallFunction {
+                        name: "echo".to_string(),
+                        arguments: r#"{"message":"hi"}"#.to_string(),
+                    },
+                }]),
+            },
+            ChatCompletionMessage {
+                role: "tool".to_string(),
+                content: Some("hi".to_string()),
+                tool_call_id: Some("call_1".to_string()),
+                name: Some("echo".to_string()),
+                tool_calls: None,
+            },
+        ];
+
+        let result = flatten_tool_messages(messages);
+        assert_eq!(result.len(), 3);
+
+        // Assistant tool_calls → plain assistant text
+        assert_eq!(result[1].role, "assistant");
+        assert!(result[1].tool_calls.is_none());
+        assert!(
+            result[1]
+                .content
+                .as_ref()
+                .unwrap()
+                .contains("[Called tool `echo`")
+        );
+
+        // Tool result → user message
+        assert_eq!(result[2].role, "user");
+        assert!(result[2].tool_call_id.is_none());
+        assert!(
+            result[2]
+                .content
+                .as_ref()
+                .unwrap()
+                .contains("[Tool `echo` returned: hi]")
+        );
+    }
+
+    #[test]
+    fn test_flatten_preserves_assistant_text_with_tool_calls() {
+        let messages = vec![
+            ChatCompletionMessage {
+                role: "assistant".to_string(),
+                content: Some("Let me check that.".to_string()),
+                tool_call_id: None,
+                name: None,
+                tool_calls: Some(vec![ChatCompletionToolCall {
+                    id: "call_1".to_string(),
+                    call_type: "function".to_string(),
+                    function: ChatCompletionToolCallFunction {
+                        name: "search".to_string(),
+                        arguments: r#"{"q":"test"}"#.to_string(),
+                    },
+                }]),
+            },
+            ChatCompletionMessage {
+                role: "tool".to_string(),
+                content: Some("found it".to_string()),
+                tool_call_id: Some("call_1".to_string()),
+                name: Some("search".to_string()),
+                tool_calls: None,
+            },
+        ];
+
+        let result = flatten_tool_messages(messages);
+        let text = result[0].content.as_ref().unwrap();
+        assert!(text.starts_with("Let me check that."));
+        assert!(text.contains("[Called tool `search`"));
+    }
 }

src/llm/reasoning.rs

@@ -113,6 +113,25 @@ pub struct ToolSelection {
     pub reasoning: String,
     /// Alternative tools considered.
     pub alternatives: Vec<String>,
+    /// The tool call ID from the LLM response.
+    ///
+    /// OpenAI-compatible providers assign each tool call a unique ID that must
+    /// be echoed back in the corresponding tool result message. Without this,
+    /// the provider cannot match results to their originating calls.
+    pub tool_call_id: String,
+}
+
+/// Token usage from a single LLM call.
+#[derive(Debug, Clone, Copy, Default)]
+pub struct TokenUsage {
+    pub input_tokens: u32,
+    pub output_tokens: u32,
+}
+
+impl TokenUsage {
+    pub fn total(&self) -> u32 {
+        self.input_tokens + self.output_tokens
+    }
 }
 
 /// Result of a response with potential tool calls.
@@ -131,6 +150,13 @@ pub enum RespondResult {
     },
 }
 
+/// A `RespondResult` bundled with the token usage from the LLM call that produced it.
+#[derive(Debug, Clone)]
+pub struct RespondOutput {
+    pub result: RespondResult,
+    pub usage: TokenUsage,
+}
+
 /// Reasoning engine for the agent.
 pub struct Reasoning {
     llm: Arc<dyn LlmProvider>,
@@ -224,6 +250,7 @@ impl Reasoning {
                 parameters: tool_call.arguments,
                 reasoning: reasoning.clone(),
                 alternatives: vec![],
+                tool_call_id: tool_call.id,
             })
             .collect();
 
@@ -284,7 +311,8 @@ Respond in JSON format:
     /// tool calls as text for simple cases. Use `respond_with_tools()` when you
     /// need to actually execute tool calls in an agentic loop.
     pub async fn respond(&self, context: &ReasoningContext) -> Result<String, LlmError> {
-        match self.respond_with_tools(context).await? {
+        let output = self.respond_with_tools(context).await?;
+        match output.result {
             RespondResult::Text(text) => Ok(text),
             RespondResult::ToolCalls {
                 tool_calls: calls, ..
@@ -299,15 +327,14 @@ Respond in JSON format:
         }
     }
 
-    /// Generate a response that may include tool calls.
+    /// Generate a response that may include tool calls, with token usage tracking.
     ///
-    /// Returns `RespondResult::ToolCalls` if the model wants to call tools,
-    /// allowing the caller to execute them and continue the conversation.
-    /// Returns `RespondResult::Text` when the model has a final text response.
+    /// Returns `RespondOutput` containing the result and token usage from the LLM call.
+    /// The caller should use `usage` to track cost/budget against the job.
     pub async fn respond_with_tools(
         &self,
         context: &ReasoningContext,
-    ) -> Result<RespondResult, LlmError> {
+    ) -> Result<RespondOutput, LlmError> {
         let system_prompt = self.build_conversation_prompt(context);
 
         let mut messages = vec![ChatMessage::system(system_prompt)];
@@ -322,12 +349,19 @@ Respond in JSON format:
             request.metadata = context.metadata.clone();
 
             let response = self.llm.complete_with_tools(request).await?;
+            let usage = TokenUsage {
+                input_tokens: response.input_tokens,
+                output_tokens: response.output_tokens,
+            };
 
             // If there were tool calls, return them for execution
             if !response.tool_calls.is_empty() {
-                return Ok(RespondResult::ToolCalls {
-                    tool_calls: response.tool_calls,
-                    content: response.content,
+                return Ok(RespondOutput {
+                    result: RespondResult::ToolCalls {
+                        tool_calls: response.tool_calls,
+                        content: response.content,
+                    },
+                    usage,
                 });
             }
 
@@ -341,17 +375,23 @@ Respond in JSON format:
             let recovered = recover_tool_calls_from_content(&content, &context.available_tools);
             if !recovered.is_empty() {
                 let cleaned = clean_response(&content);
-                return Ok(RespondResult::ToolCalls {
-                    tool_calls: recovered,
-                    content: if cleaned.is_empty() {
-                        None
-                    } else {
-                        Some(cleaned)
+                return Ok(RespondOutput {
+                    result: RespondResult::ToolCalls {
+                        tool_calls: recovered,
+                        content: if cleaned.is_empty() {
+                            None
+                        } else {
+                            Some(cleaned)
+                        },
                     },
+                    usage,
                 });
             }
 
-            Ok(RespondResult::Text(clean_response(&content)))
+            Ok(RespondOutput {
+                result: RespondResult::Text(clean_response(&content)),
+                usage,
+            })
         } else {
             // No tools, use simple completion
             let mut request = CompletionRequest::new(messages)
@@ -360,7 +400,13 @@ Respond in JSON format:
             request.metadata = context.metadata.clone();
 
             let response = self.llm.complete(request).await?;
-            Ok(RespondResult::Text(clean_response(&response.content)))
+            Ok(RespondOutput {
+                result: RespondResult::Text(clean_response(&response.content)),
+                usage: TokenUsage {
+                    input_tokens: response.input_tokens,
+                    output_tokens: response.output_tokens,
+                },
+            })
         }
     }
 
@@ -542,21 +588,20 @@ fn recover_tool_calls_from_content(
             }
 
             // Try JSON first: {"name":"x","arguments":{}}
-            if let Ok(parsed) = serde_json::from_str::<serde_json::Value>(inner) {
-                if let Some(name) = parsed.get("name").and_then(|v| v.as_str()) {
-                    if tool_names.contains(name) {
-                        let arguments = parsed
-                            .get("arguments")
-                            .cloned()
-                            .unwrap_or(serde_json::Value::Object(Default::default()));
-                        calls.push(ToolCall {
-                            id: format!("recovered_{}", calls.len()),
-                            name: name.to_string(),
-                            arguments,
-                        });
-                        continue;
-                    }
-                }
+            if let Ok(parsed) = serde_json::from_str::<serde_json::Value>(inner)
+                && let Some(name) = parsed.get("name").and_then(|v| v.as_str())
+                && tool_names.contains(name)
+            {
+                let arguments = parsed
+                    .get("arguments")
+                    .cloned()
+                    .unwrap_or(serde_json::Value::Object(Default::default()));
+                calls.push(ToolCall {
+                    id: format!("recovered_{}", calls.len()),
+                    name: name.to_string(),
+                    arguments,
+                });
+                continue;
             }
 
             // Bare tool name (e.g. "<tool_call>tool_list</tool_call>")

src/llm/retry.rs

@@ -0,0 +1,96 @@
+//! Shared retry helpers for LLM providers.
+//!
+//! Provides exponential backoff with jitter and retryable status classification
+//! used by both `NearAiProvider` and `NearAiChatProvider`.
+
+use std::time::Duration;
+
+use rand::Rng;
+
+/// Returns `true` if the HTTP status code is transient and worth retrying.
+pub(crate) fn is_retryable_status(status: u16) -> bool {
+    matches!(status, 429 | 500 | 502 | 503 | 504)
+}
+
+/// Calculate exponential backoff delay with random jitter.
+///
+/// Base delay is 1 second, doubled each attempt, with +/-25% jitter.
+/// - attempt 0: ~1s (0.75s - 1.25s)
+/// - attempt 1: ~2s (1.5s - 2.5s)
+/// - attempt 2: ~4s (3.0s - 5.0s)
+pub(crate) fn retry_backoff_delay(attempt: u32) -> Duration {
+    let base_ms: u64 = 1000u64.saturating_mul(2u64.saturating_pow(attempt));
+    let jitter_range = base_ms / 4; // 25%
+    let jitter = if jitter_range > 0 {
+        let offset = rand::thread_rng().gen_range(0..=jitter_range * 2);
+        offset as i64 - jitter_range as i64
+    } else {
+        0
+    };
+    let delay_ms = (base_ms as i64 + jitter).max(100) as u64;
+    Duration::from_millis(delay_ms)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_is_retryable_status() {
+        // Transient errors should be retryable
+        assert!(is_retryable_status(429));
+        assert!(is_retryable_status(500));
+        assert!(is_retryable_status(502));
+        assert!(is_retryable_status(503));
+        assert!(is_retryable_status(504));
+
+        // Client errors should not be retryable
+        assert!(!is_retryable_status(400));
+        assert!(!is_retryable_status(401));
+        assert!(!is_retryable_status(403));
+        assert!(!is_retryable_status(404));
+        assert!(!is_retryable_status(422));
+
+        // Success codes should not be retryable
+        assert!(!is_retryable_status(200));
+        assert!(!is_retryable_status(201));
+    }
+
+    #[test]
+    fn test_retry_backoff_delay_exponential_growth() {
+        // Run multiple samples to verify the range, accounting for jitter
+        for _ in 0..20 {
+            let d0 = retry_backoff_delay(0);
+            let d1 = retry_backoff_delay(1);
+            let d2 = retry_backoff_delay(2);
+
+            // Attempt 0: base 1000ms, jitter +/-250ms -> [750, 1250]
+            assert!(d0.as_millis() >= 750, "attempt 0 too low: {:?}", d0);
+            assert!(d0.as_millis() <= 1250, "attempt 0 too high: {:?}", d0);
+
+            // Attempt 1: base 2000ms, jitter +/-500ms -> [1500, 2500]
+            assert!(d1.as_millis() >= 1500, "attempt 1 too low: {:?}", d1);
+            assert!(d1.as_millis() <= 2500, "attempt 1 too high: {:?}", d1);
+
+            // Attempt 2: base 4000ms, jitter +/-1000ms -> [3000, 5000]
+            assert!(d2.as_millis() >= 3000, "attempt 2 too low: {:?}", d2);
+            assert!(d2.as_millis() <= 5000, "attempt 2 too high: {:?}", d2);
+        }
+    }
+
+    #[test]
+    fn test_retry_backoff_delay_minimum() {
+        // Even at attempt 0, delay should be at least 100ms (the minimum floor)
+        for _ in 0..20 {
+            let delay = retry_backoff_delay(0);
+            assert!(delay.as_millis() >= 100);
+        }
+    }
+
+    #[test]
+    fn test_retry_backoff_delay_no_overflow() {
+        // Very high attempt numbers should not panic from overflow
+        let delay = retry_backoff_delay(30);
+        assert!(delay.as_millis() >= 100);
+    }
+}

src/llm/session.rs

@@ -31,8 +31,6 @@ pub struct SessionConfig {
     pub auth_base_url: String,
     /// Path to session file (e.g., ~/.ironclaw/session.json).
     pub session_path: PathBuf,
-    /// Port range for OAuth callback server.
-    pub callback_port_range: (u16, u16),
 }
 
 impl Default for SessionConfig {
@@ -40,7 +38,6 @@ impl Default for SessionConfig {
         Self {
             auth_base_url: "https://private.near.ai".to_string(),
             session_path: default_session_path(),
-            callback_port_range: (9876, 9886),
         }
     }
 }
@@ -62,7 +59,7 @@ pub struct SessionManager {
     /// Prevents thundering herd during concurrent 401s.
     renewal_lock: Mutex<()>,
     /// Optional database store for persisting session to the settings table.
-    store: RwLock<Option<Arc<crate::history::Store>>>,
+    store: RwLock<Option<Arc<dyn crate::db::Database>>>,
     /// User ID for DB settings (default: "default").
     user_id: RwLock<String>,
 }
@@ -83,16 +80,16 @@ impl SessionManager {
         };
 
         // Try to load existing session synchronously during construction
-        if let Ok(data) = std::fs::read_to_string(&manager.config.session_path) {
-            if let Ok(session) = serde_json::from_str::<SessionData>(&data) {
-                // We can't await here, so we use try_write
-                if let Ok(mut guard) = manager.token.try_write() {
-                    *guard = Some(SecretString::from(session.session_token));
-                    tracing::info!(
-                        "Loaded session token from {}",
-                        manager.config.session_path.display()
-                    );
-                }
+        if let Ok(data) = std::fs::read_to_string(&manager.config.session_path)
+            && let Ok(session) = serde_json::from_str::<SessionData>(&data)
+        {
+            // We can't await here, so we use try_write
+            if let Ok(mut guard) = manager.token.try_write() {
+                *guard = Some(SecretString::from(session.session_token));
+                tracing::info!(
+                    "Loaded session token from {}",
+                    manager.config.session_path.display()
+                );
             }
         }
 
@@ -125,7 +122,7 @@ impl SessionManager {
     /// When a store is attached, session tokens are saved to the `settings`
     /// table (key: `nearai.session_token`) in addition to the disk file.
     /// On load, DB is preferred over disk.
-    pub async fn attach_store(&self, store: Arc<crate::history::Store>, user_id: &str) {
+    pub async fn attach_store(&self, store: Arc<dyn crate::db::Database>, user_id: &str) {
         *self.store.write().await = Some(store);
         *self.user_id.write().await = user_id.to_string();
 
@@ -222,38 +219,21 @@ impl SessionManager {
 
     /// Start the OAuth login flow.
     ///
-    /// 1. Find an available port for the callback server
+    /// 1. Bind the fixed callback port
     /// 2. Print the auth URL and attempt to open browser
     /// 3. Wait for OAuth callback with session token
     /// 4. Save and return the token
     async fn initiate_login(&self) -> Result<(), LlmError> {
-        use tokio::io::{AsyncBufReadExt, AsyncWriteExt, BufReader};
-        use tokio::net::TcpListener;
+        use crate::cli::oauth_defaults::{self, OAUTH_CALLBACK_PORT};
 
-        // Find an available port
-        let mut listener = None;
-        let mut port = 0;
+        let listener = oauth_defaults::bind_callback_listener()
+            .await
+            .map_err(|e| LlmError::SessionRenewalFailed {
+                provider: "nearai".to_string(),
+                reason: e.to_string(),
+            })?;
 
-        for p in self.config.callback_port_range.0..=self.config.callback_port_range.1 {
-            match TcpListener::bind(format!("127.0.0.1:{}", p)).await {
-                Ok(l) => {
-                    listener = Some(l);
-                    port = p;
-                    break;
-                }
-                Err(_) => continue,
-            }
-        }
-
-        let listener = listener.ok_or_else(|| LlmError::SessionRenewalFailed {
-            provider: "nearai".to_string(),
-            reason: format!(
-                "Could not find available port in range {}-{}",
-                self.config.callback_port_range.0, self.config.callback_port_range.1
-            ),
-        })?;
-
-        let callback_url = format!("http://127.0.0.1:{}", port);
+        let callback_url = format!("http://127.0.0.1:{}", OAUTH_CALLBACK_PORT);
 
         // Show auth provider menu
         println!();
@@ -333,138 +313,16 @@ impl SessionManager {
         println!();
         println!("Waiting for authentication...");
 
-        // Wait for callback with timeout
-        // The API redirects to: {frontend_callback}/auth/callback?token=X&session_id=X&expires_at=X&is_new_user=X
-        let timeout = std::time::Duration::from_secs(300); // 5 minutes
-        let selected_provider = auth_provider.to_string();
-        let (session_token, auth_provider) = tokio::time::timeout(timeout, async move {
-            loop {
-                let (mut socket, _) = listener.accept().await.map_err(|e| {
-                    LlmError::SessionRenewalFailed {
-                        provider: "nearai".to_string(),
-                        reason: format!("Failed to accept connection: {}", e),
-                    }
+        // The NEAR AI API redirects to: {frontend_callback}/auth/callback?token=X&...
+        let session_token =
+            oauth_defaults::wait_for_callback(listener, "/auth/callback", "token", "NEAR AI")
+                .await
+                .map_err(|e| LlmError::SessionRenewalFailed {
+                    provider: "nearai".to_string(),
+                    reason: e.to_string(),
                 })?;
 
-                let mut reader = BufReader::new(&mut socket);
-                let mut request_line = String::new();
-                reader.read_line(&mut request_line).await.map_err(|e| {
-                    LlmError::SessionRenewalFailed {
-                        provider: "nearai".to_string(),
-                        reason: format!("Failed to read request: {}", e),
-                    }
-                })?;
-
-                // Parse GET /auth/callback?token=xxx&session_id=xxx&expires_at=xxx&is_new_user=xxx HTTP/1.1
-                if let Some(path) = request_line.split_whitespace().nth(1) {
-                    if path.starts_with("/auth/callback") {
-                        // Parse query parameters
-                        if let Some(query) = path.split('?').nth(1) {
-                            let mut token = None;
-
-                            for param in query.split('&') {
-                                let parts: Vec<&str> = param.splitn(2, '=').collect();
-                                if parts.len() == 2 && parts[0] == "token" {
-                                    token = Some(
-                                        urlencoding::decode(parts[1])
-                                            .unwrap_or_else(|_| parts[1].into())
-                                            .into_owned(),
-                                    );
-                                }
-                            }
-
-                            if let Some(token) = token {
-                                // Send success response with nice styling
-                                let response = concat!(
-                                    "HTTP/1.1 200 OK\r\n",
-                                    "Content-Type: text/html; charset=utf-8\r\n",
-                                    "Connection: close\r\n",
-                                    "\r\n",
-                                    "<!DOCTYPE html>\n",
-                                    "<html>\n",
-                                    "<head>\n",
-                                    "  <meta charset=\"utf-8\">\n",
-                                    "  <title>NEAR AI - Authentication Successful</title>\n",
-                                    "  <style>\n",
-                                    "    * { margin: 0; padding: 0; box-sizing: border-box; }\n",
-                                    "    body {\n",
-                                    "      font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, sans-serif;\n",
-                                    "      background: linear-gradient(135deg, #1a1a2e 0%, #16213e 100%);\n",
-                                    "      min-height: 100vh;\n",
-                                    "      display: flex;\n",
-                                    "      align-items: center;\n",
-                                    "      justify-content: center;\n",
-                                    "      color: #fff;\n",
-                                    "    }\n",
-                                    "    .container {\n",
-                                    "      text-align: center;\n",
-                                    "      padding: 3rem;\n",
-                                    "      background: rgba(255,255,255,0.05);\n",
-                                    "      border-radius: 16px;\n",
-                                    "      backdrop-filter: blur(10px);\n",
-                                    "      border: 1px solid rgba(255,255,255,0.1);\n",
-                                    "      max-width: 400px;\n",
-                                    "    }\n",
-                                    "    .checkmark {\n",
-                                    "      width: 80px;\n",
-                                    "      height: 80px;\n",
-                                    "      background: linear-gradient(135deg, #00d9a5 0%, #00b386 100%);\n",
-                                    "      border-radius: 50%;\n",
-                                    "      display: flex;\n",
-                                    "      align-items: center;\n",
-                                    "      justify-content: center;\n",
-                                    "      margin: 0 auto 1.5rem;\n",
-                                    "      font-size: 40px;\n",
-                                    "    }\n",
-                                    "    h1 {\n",
-                                    "      font-size: 1.5rem;\n",
-                                    "      font-weight: 600;\n",
-                                    "      margin-bottom: 0.75rem;\n",
-                                    "    }\n",
-                                    "    p {\n",
-                                    "      color: rgba(255,255,255,0.7);\n",
-                                    "      font-size: 0.95rem;\n",
-                                    "      line-height: 1.5;\n",
-                                    "    }\n",
-                                    "    .brand {\n",
-                                    "      margin-top: 2rem;\n",
-                                    "      padding-top: 1.5rem;\n",
-                                    "      border-top: 1px solid rgba(255,255,255,0.1);\n",
-                                    "      font-size: 0.8rem;\n",
-                                    "      color: rgba(255,255,255,0.4);\n",
-                                    "    }\n",
-                                    "  </style>\n",
-                                    "</head>\n",
-                                    "<body>\n",
-                                    "  <div class=\"container\">\n",
-                                    "    <div class=\"checkmark\">&#10003;</div>\n",
-                                    "    <h1>Authentication Successful</h1>\n",
-                                    "    <p>You can close this window and return to the terminal.</p>\n",
-                                    "    <div class=\"brand\">NEAR AI Agent</div>\n",
-                                    "  </div>\n",
-                                    "</body>\n",
-                                    "</html>"
-                                );
-
-                                let _ = socket.write_all(response.as_bytes()).await;
-                                let _ = socket.shutdown().await;
-
-                                return Ok::<_, LlmError>((token, Some(selected_provider.clone())));
-                            }
-                        }
-                    }
-                }
-
-                // Not the callback we're looking for, send 404
-                let response = "HTTP/1.1 404 Not Found\r\nConnection: close\r\n\r\n";
-                let _ = socket.write_all(response.as_bytes()).await;
-            }
-        })
-        .await
-        .map_err(|_| LlmError::SessionRenewalFailed {
-            provider: "nearai".to_string(),
-            reason: "Authentication timed out after 5 minutes".to_string(),
-        })??;
+        let auth_provider = Some(auth_provider.to_string());
 
         // Save the token
         self.save_session(&session_token, auth_provider.as_deref())
@@ -520,6 +378,25 @@ impl SessionManager {
                 ))
             })?;
 
+        // Restrictive permissions: session file contains a secret token
+        #[cfg(unix)]
+        {
+            use std::os::unix::fs::PermissionsExt;
+            let perms = std::fs::Permissions::from_mode(0o600);
+            tokio::fs::set_permissions(&self.config.session_path, perms)
+                .await
+                .map_err(|e| {
+                    LlmError::Io(std::io::Error::new(
+                        e.kind(),
+                        format!(
+                            "Failed to set permissions on {}: {}",
+                            self.config.session_path.display(),
+                            e
+                        ),
+                    ))
+                })?;
+        }
+
         tracing::debug!("Session saved to {}", self.config.session_path.display());
 
         // Also save to DB if a store is attached
@@ -551,17 +428,30 @@ impl SessionManager {
             })?;
 
         let user_id = self.user_id.read().await.clone();
-        let value = store
+        let value = if let Some(value) = store
             .get_setting(&user_id, "nearai.session_token")
             .await
             .map_err(|e| LlmError::SessionRenewalFailed {
                 provider: "nearai".to_string(),
                 reason: format!("DB query failed: {}", e),
-            })?
-            .ok_or_else(|| LlmError::SessionRenewalFailed {
-                provider: "nearai".to_string(),
-                reason: "No session in DB".to_string(),
-            })?;
+            })? {
+            value
+        } else {
+            tracing::warn!(
+                "nearai.session_token missing; falling back to legacy nearai.session for backwards compatibility"
+            );
+            store
+                .get_setting(&user_id, "nearai.session")
+                .await
+                .map_err(|e| LlmError::SessionRenewalFailed {
+                    provider: "nearai".to_string(),
+                    reason: format!("DB query failed: {}", e),
+                })?
+                .ok_or(LlmError::SessionRenewalFailed {
+                    provider: "nearai".to_string(),
+                    reason: "No session in DB".to_string(),
+                })?
+        };
 
         let session: SessionData =
             serde_json::from_value(value).map_err(|e| LlmError::SessionRenewalFailed {
@@ -623,15 +513,14 @@ pub async fn create_session_manager(config: SessionConfig) -> Arc<SessionManager
     let manager = SessionManager::new_async(config).await;
 
     // Check for legacy env var and migrate if present and no file token
-    if !manager.has_token().await {
-        if let Ok(token) = std::env::var("NEARAI_SESSION_TOKEN") {
-            if !token.is_empty() {
-                tracing::info!("Migrating session token from NEARAI_SESSION_TOKEN env var to file");
-                manager.set_token(SecretString::from(token.clone())).await;
-                if let Err(e) = manager.save_session(&token, None).await {
-                    tracing::warn!("Failed to save migrated session: {}", e);
-                }
-            }
+    if !manager.has_token().await
+        && let Ok(token) = std::env::var("NEARAI_SESSION_TOKEN")
+        && !token.is_empty()
+    {
+        tracing::info!("Migrating session token from NEARAI_SESSION_TOKEN env var to file");
+        manager.set_token(SecretString::from(token.clone())).await;
+        if let Err(e) = manager.save_session(&token, None).await {
+            tracing::warn!("Failed to save migrated session: {}", e);
         }
     }
 
@@ -652,7 +541,6 @@ mod tests {
         let config = SessionConfig {
             auth_base_url: "https://example.com".to_string(),
             session_path: session_path.clone(),
-            callback_port_range: (9900, 9910),
         };
 
         let manager = SessionManager::new_async(config.clone()).await;
@@ -693,7 +581,6 @@ mod tests {
         let config = SessionConfig {
             auth_base_url: "https://example.com".to_string(),
             session_path: dir.path().join("nonexistent.json"),
-            callback_port_range: (9900, 9910),
         };
 
         let manager = SessionManager::new_async(config).await;

src/main.rs

@@ -17,22 +17,23 @@ use ironclaw::{
         web::log_layer::{LogBroadcaster, WebLogLayer},
     },
     cli::{
-        Cli, Command, run_mcp_command, run_memory_command, run_pairing_command, run_status_command,
-        run_tool_command,
+        Cli, Command, run_mcp_command, run_pairing_command, run_status_command, run_tool_command,
     },
     config::Config,
     context::ContextManager,
     extensions::ExtensionManager,
-    history::Store,
-    llm::{SessionConfig, create_llm_provider, create_session_manager},
+    hooks::HookRegistry,
+    llm::{
+        CooldownConfig, FailoverProvider, LlmProvider, SessionConfig, create_cheap_llm_provider,
+        create_llm_provider, create_llm_provider_with_config, create_session_manager,
+    },
     orchestrator::{
         ContainerJobConfig, ContainerJobManager, OrchestratorApi, TokenStore,
         api::OrchestratorState,
     },
     pairing::PairingStore,
     safety::SafetyLayer,
-    secrets::{PostgresSecretsStore, SecretsCrypto, SecretsStore},
-    setup::{SetupConfig, SetupWizard},
+    secrets::SecretsStore,
     tools::{
         ToolRegistry,
         mcp::{McpClient, McpSessionManager, config::load_mcp_servers_from_db, is_authenticated},
@@ -41,6 +42,13 @@ use ironclaw::{
     workspace::{EmbeddingProvider, NearAiEmbeddings, OpenAiEmbeddings, Workspace},
 };
 
+#[cfg(feature = "libsql")]
+use ironclaw::secrets::LibSqlSecretsStore;
+#[cfg(feature = "postgres")]
+use ironclaw::secrets::PostgresSecretsStore;
+use ironclaw::secrets::SecretsCrypto;
+#[cfg(any(feature = "postgres", feature = "libsql"))]
+use ironclaw::setup::{SetupConfig, SetupWizard};
 #[tokio::main]
 async fn main() -> anyhow::Result<()> {
     let cli = Cli::parse();
@@ -85,18 +93,14 @@ async fn main() -> anyhow::Result<()> {
                 .init();
 
             // Memory commands need database (and optionally embeddings)
-            let _ = dotenvy::dotenv();
             let config = Config::from_env()
                 .await
                 .map_err(|e| anyhow::anyhow!("{}", e))?;
-            let store = ironclaw::history::Store::new(&config.database).await?;
-            store.run_migrations().await?;
 
             // Set up embeddings if available
             let session = ironclaw::llm::create_session_manager(ironclaw::llm::SessionConfig {
                 auth_base_url: config.llm.nearai.auth_base_url.clone(),
                 session_path: config.llm.nearai.session_path.clone(),
-                ..Default::default()
             })
             .await;
 
@@ -130,7 +134,14 @@ async fn main() -> anyhow::Result<()> {
                     None
                 };
 
-            return run_memory_command(mem_cmd.clone(), store.pool(), embeddings).await;
+            // Create a Database-trait-backed workspace for the memory command
+            let db: Arc<dyn ironclaw::db::Database> =
+                ironclaw::db::connect_from_config(&config.database)
+                    .await
+                    .map_err(|e| anyhow::anyhow!("{}", e))?;
+
+            return ironclaw::cli::run_memory_command_with_db(mem_cmd.clone(), db, embeddings)
+                .await;
         }
         Some(Command::Pairing(pairing_cmd)) => {
             tracing_subscriber::fmt()
@@ -142,7 +153,6 @@ async fn main() -> anyhow::Result<()> {
             return run_pairing_command(pairing_cmd.clone()).map_err(|e| anyhow::anyhow!("{}", e));
         }
         Some(Command::Status) => {
-            let _ = dotenvy::dotenv();
             tracing_subscriber::fmt()
                 .with_env_filter(
                     EnvFilter::try_from_default_env().unwrap_or_else(|_| EnvFilter::new("warn")),
@@ -216,6 +226,7 @@ async fn main() -> anyhow::Result<()> {
                 max_turns: *max_turns,
                 model: model.clone(),
                 timeout: std::time::Duration::from_secs(1800),
+                allowed_tools: Vec::new(),
             };
 
             let runtime = ironclaw::worker::ClaudeBridgeRuntime::new(config)
@@ -232,15 +243,25 @@ async fn main() -> anyhow::Result<()> {
             skip_auth,
             channels_only,
         }) => {
-            // Load .env before running onboarding wizard
+            // Load .env files before running onboarding wizard.
+            // Standard ./.env first (higher priority), then ~/.ironclaw/.env.
             let _ = dotenvy::dotenv();
+            ironclaw::bootstrap::load_ironclaw_env();
 
-            let config = SetupConfig {
-                skip_auth: *skip_auth,
-                channels_only: *channels_only,
-            };
-            let mut wizard = SetupWizard::with_config(config);
-            wizard.run().await?;
+            #[cfg(any(feature = "postgres", feature = "libsql"))]
+            {
+                let config = SetupConfig {
+                    skip_auth: *skip_auth,
+                    channels_only: *channels_only,
+                };
+                let mut wizard = SetupWizard::with_config(config);
+                wizard.run().await?;
+            }
+            #[cfg(not(any(feature = "postgres", feature = "libsql")))]
+            {
+                let _ = (skip_auth, channels_only);
+                eprintln!("Onboarding wizard requires the 'postgres' or 'libsql' feature.");
+            }
             return Ok(());
         }
         None | Some(Command::Run) => {
@@ -248,22 +269,23 @@ async fn main() -> anyhow::Result<()> {
         }
     }
 
-    // Load .env if present
+    // Load .env files early so DATABASE_URL (and any other vars) are
+    // available to all subsequent env-based config resolution.
+    // Standard ./.env first (higher priority), then ~/.ironclaw/.env.
     let _ = dotenvy::dotenv();
+    ironclaw::bootstrap::load_ironclaw_env();
 
     // Enhanced first-run detection
-    if !cli.no_onboard {
-        if let Some(reason) = check_onboard_needed().await {
-            println!("Onboarding needed: {}", reason);
-            println!();
-            let mut wizard = SetupWizard::new();
-            wizard.run().await?;
-        }
+    #[cfg(any(feature = "postgres", feature = "libsql"))]
+    if !cli.no_onboard
+        && let Some(reason) = check_onboard_needed()
+    {
+        println!("Onboarding needed: {}", reason);
+        println!();
+        let mut wizard = SetupWizard::new();
+        wizard.run().await?;
     }
 
-    // Load bootstrap config (4 fields that must live on disk)
-    let bootstrap = ironclaw::bootstrap::BootstrapConfig::load();
-
     // Load initial config from env + disk (before DB is available)
     let mut config = match Config::from_env().await {
         Ok(c) => c,
@@ -283,18 +305,20 @@ async fn main() -> anyhow::Result<()> {
     let session_config = SessionConfig {
         auth_base_url: config.llm.nearai.auth_base_url.clone(),
         session_path: config.llm.nearai.session_path.clone(),
-        ..Default::default()
     };
     let session = create_session_manager(session_config).await;
 
-    // Ensure we're authenticated before proceeding (only needed for NEAR AI backend)
-    if config.llm.backend == ironclaw::config::LlmBackend::NearAi {
+    // Session-based auth is only needed for NEAR AI backend without an API key.
+    // ChatCompletions mode with an API key skips session auth entirely.
+    if config.llm.backend == ironclaw::config::LlmBackend::NearAi
+        && config.llm.nearai.api_key.is_none()
+    {
         session.ensure_authenticated().await?;
     }
 
     // Initialize tracing
     let env_filter = EnvFilter::try_from_default_env()
-        .unwrap_or_else(|_| EnvFilter::new("ironclaw=info,tower_http=debug"));
+        .unwrap_or_else(|_| EnvFilter::new("ironclaw=info,tower_http=warn"));
 
     // Create log broadcaster before tracing init so the WebLogLayer can capture all events.
     // This gets wired to the gateway's /api/logs/events SSE endpoint later.
@@ -302,7 +326,11 @@ async fn main() -> anyhow::Result<()> {
 
     tracing_subscriber::registry()
         .with(env_filter)
-        .with(tracing_subscriber::fmt::layer().with_target(false))
+        .with(
+            tracing_subscriber::fmt::layer()
+                .with_target(false)
+                .with_writer(ironclaw::tracing_fmt::TruncatingStderr::default()),
+        )
         .with(WebLogLayer::new(Arc::clone(&log_broadcaster)))
         .init();
 
@@ -310,7 +338,10 @@ async fn main() -> anyhow::Result<()> {
     let repl_channel = if let Some(ref msg) = cli.message {
         Some(ReplChannel::with_message(msg.clone()))
     } else if config.channels.cli.enabled {
-        Some(ReplChannel::new())
+        let repl = ReplChannel::new();
+        // Suppress the one-liner banner; boot screen will be shown instead.
+        repl.suppress_banner();
+        Some(repl)
     } else {
         None
     };
@@ -319,23 +350,86 @@ async fn main() -> anyhow::Result<()> {
     tracing::info!("Loaded configuration for agent: {}", config.agent.name);
     tracing::info!("LLM backend: {}", config.llm.backend);
 
-    // Initialize database store (optional for testing)
-    let store = if cli.no_db {
+    // Initialize database backend.
+    //
+    // Creates an `Arc<dyn Database>` that all consumers share.
+    // Backend is selected by the `DATABASE_BACKEND` env var / config.
+    //
+    // NOTE: For simpler call sites (CLI commands, Memory handler) use the shared
+    // helper `ironclaw::db::connect_from_config()`. This block is kept inline
+    // because it also captures backend-specific handles (`pg_pool`, `libsql_db`)
+    // needed by the secrets store.
+    #[cfg(feature = "postgres")]
+    let mut pg_pool: Option<deadpool_postgres::Pool> = None;
+    #[cfg(feature = "libsql")]
+    let mut libsql_db: Option<std::sync::Arc<libsql::Database>> = None;
+
+    let db: Option<Arc<dyn ironclaw::db::Database>> = if cli.no_db {
         tracing::warn!("Running without database connection");
         None
     } else {
-        let store = Store::new(&config.database).await?;
-        store.run_migrations().await?;
-        tracing::info!("Database connected and migrations applied");
+        match config.database.backend {
+            #[cfg(feature = "libsql")]
+            ironclaw::config::DatabaseBackend::LibSql => {
+                use ironclaw::db::Database as _;
+                use ironclaw::db::libsql_backend::LibSqlBackend;
+                use secrecy::ExposeSecret as _;
 
+                let default_path = ironclaw::config::default_libsql_path();
+                let db_path = config
+                    .database
+                    .libsql_path
+                    .as_deref()
+                    .unwrap_or(&default_path);
+
+                let backend = if let Some(ref url) = config.database.libsql_url {
+                    let token = config.database.libsql_auth_token.as_ref().ok_or_else(|| {
+                        anyhow::anyhow!("LIBSQL_AUTH_TOKEN is required when LIBSQL_URL is set")
+                    })?;
+                    LibSqlBackend::new_remote_replica(db_path, url, token.expose_secret()).await?
+                } else {
+                    LibSqlBackend::new_local(db_path).await?
+                };
+                backend.run_migrations().await?;
+                tracing::info!("libSQL database connected and migrations applied");
+
+                // Capture the Database handle for SecretsStore (connection-per-op)
+                libsql_db = Some(backend.shared_db());
+
+                Some(Arc::new(backend) as Arc<dyn ironclaw::db::Database>)
+            }
+            #[cfg(feature = "postgres")]
+            _ => {
+                use ironclaw::db::Database as _;
+                let pg = ironclaw::db::postgres::PgBackend::new(&config.database)
+                    .await
+                    .map_err(|e| anyhow::anyhow!("{}", e))?;
+                pg.run_migrations()
+                    .await
+                    .map_err(|e| anyhow::anyhow!("{}", e))?;
+                tracing::info!("PostgreSQL database connected and migrations applied");
+
+                pg_pool = Some(pg.pool());
+                Some(Arc::new(pg) as Arc<dyn ironclaw::db::Database>)
+            }
+            #[cfg(not(feature = "postgres"))]
+            _ => {
+                anyhow::bail!(
+                    "No database backend available. Enable 'postgres' or 'libsql' feature."
+                );
+            }
+        }
+    };
+
+    // Post-init operations using the database
+    if let Some(ref db) = db {
         // One-time migration: move disk config files into the DB settings table.
-        if let Err(e) = ironclaw::bootstrap::migrate_disk_to_db(&store, "default").await {
+        if let Err(e) = ironclaw::bootstrap::migrate_disk_to_db(db.as_ref(), "default").await {
             tracing::warn!("Disk-to-DB settings migration failed: {}", e);
         }
 
         // Reload config from DB now that we have a connection.
-        // Priority: env var > DB setting > default.
-        match Config::from_db(&store, "default", &bootstrap).await {
+        match Config::from_db(db.as_ref(), "default").await {
             Ok(db_config) => {
                 config = db_config;
                 tracing::info!("Configuration reloaded from database");
@@ -348,23 +442,120 @@ async fn main() -> anyhow::Result<()> {
             }
         }
 
-        let store = Arc::new(store);
-
-        // Attach store to session manager so tokens save to DB too
-        session.attach_store(Arc::clone(&store), "default").await;
+        // Attach DB to session manager so tokens save to DB too
+        session.attach_store(Arc::clone(db), "default").await;
 
         // Mark any jobs left in "running" or "creating" state as "interrupted".
-        if let Err(e) = store.cleanup_stale_sandbox_jobs().await {
+        if let Err(e) = db.cleanup_stale_sandbox_jobs().await {
             tracing::warn!("Failed to cleanup stale sandbox jobs: {}", e);
         }
+    }
 
-        Some(store)
-    };
+    // Create secrets store early: needed for injecting LLM API keys from encrypted
+    // storage before creating the LLM provider, and later for MCP auth + WASM channels.
+    //
+    // When both `postgres` and `libsql` features are compiled, the runtime-selected
+    // backend determines which store is created: whichever DB init branch ran will
+    // have set its handle (pg_pool or libsql_db), and the or_else chain picks it up.
+    let secrets_store: Option<Arc<dyn SecretsStore + Send + Sync>> =
+        if let Some(master_key) = config.secrets.master_key() {
+            match SecretsCrypto::new(master_key.clone()) {
+                Ok(crypto) => {
+                    let crypto = Arc::new(crypto);
+                    let store: Option<Arc<dyn SecretsStore + Send + Sync>> = None;
+
+                    #[cfg(feature = "libsql")]
+                    let store = store.or_else(|| {
+                        libsql_db.take().map(|db| {
+                            Arc::new(LibSqlSecretsStore::new(db, Arc::clone(&crypto)))
+                                as Arc<dyn SecretsStore + Send + Sync>
+                        })
+                    });
+
+                    #[cfg(feature = "postgres")]
+                    let store = store.or_else(|| {
+                        pg_pool.as_ref().map(|pool| {
+                            Arc::new(PostgresSecretsStore::new(pool.clone(), Arc::clone(&crypto)))
+                                as Arc<dyn SecretsStore + Send + Sync>
+                        })
+                    });
+
+                    store
+                }
+                Err(e) => {
+                    tracing::warn!("Failed to initialize secrets crypto: {}", e);
+                    #[cfg(feature = "libsql")]
+                    let _ = libsql_db.take();
+                    None
+                }
+            }
+        } else {
+            #[cfg(feature = "libsql")]
+            let _ = libsql_db.take();
+            None
+        };
+
+    // Inject LLM API keys from the encrypted secrets store into a thread-safe
+    // overlay so that optional_env() (used by LlmConfig::resolve()) picks them
+    // up. Then re-resolve LlmConfig with the newly available keys (backend may
+    // have been set during onboarding but the API key is in the secrets store).
+    if let Some(ref secrets) = secrets_store {
+        ironclaw::config::inject_llm_keys_from_secrets(secrets.as_ref(), "default").await;
+
+        // Re-resolve LlmConfig now that secrets overlay has been populated
+        if let Some(ref db_ref) = db {
+            match Config::from_db(db_ref.as_ref(), "default").await {
+                Ok(refreshed) => {
+                    config = refreshed;
+                    tracing::debug!("LlmConfig re-resolved after secret injection");
+                }
+                Err(e) => {
+                    tracing::warn!("Failed to re-resolve config after secret injection: {}", e);
+                }
+            }
+        }
+    }
 
     // Initialize LLM provider (clone session so we can reuse it for embeddings)
     let llm = create_llm_provider(&config.llm, session.clone())?;
     tracing::info!("LLM provider initialized: {}", llm.model_name());
 
+    // Wrap in failover if a fallback model is configured
+    let llm: Arc<dyn LlmProvider> =
+        if let Some(fallback_model) = config.llm.nearai.fallback_model.as_ref() {
+            if fallback_model == &config.llm.nearai.model {
+                tracing::warn!(
+                    "fallback_model is the same as primary model, failover may not be effective"
+                );
+            }
+            let mut fallback_config = config.llm.nearai.clone();
+            fallback_config.model = fallback_model.clone();
+            let fallback = create_llm_provider_with_config(&fallback_config, session.clone())?;
+            tracing::info!(
+                primary = %llm.model_name(),
+                fallback = %fallback.model_name(),
+                "LLM failover enabled"
+            );
+            let cooldown_config = CooldownConfig {
+                cooldown_duration: std::time::Duration::from_secs(
+                    config.llm.nearai.failover_cooldown_secs,
+                ),
+                failure_threshold: config.llm.nearai.failover_cooldown_threshold,
+            };
+            Arc::new(FailoverProvider::with_cooldown(
+                vec![llm, fallback],
+                cooldown_config,
+            )?)
+        } else {
+            llm
+        };
+
+    // Initialize cheap LLM provider for lightweight tasks (heartbeat, evaluation)
+    let cheap_llm = create_cheap_llm_provider(&config.llm, session.clone())?;
+    if let Some(ref cheap) = cheap_llm {
+        tracing::info!("Cheap LLM provider initialized: {}", cheap.model_name());
+    }
+
     // Initialize safety layer
     let safety = Arc::new(SafetyLayer::new(&config.safety));
     tracing::info!("Safety layer initialized");
@@ -414,8 +605,8 @@ async fn main() -> anyhow::Result<()> {
     };
 
     // Register memory tools if database is available
-    if let Some(ref store) = store {
-        let mut workspace = Workspace::new("default", store.pool());
+    if let Some(ref db) = db {
+        let mut workspace = Workspace::new_with_db("default", Arc::clone(db));
         if let Some(ref emb) = embeddings {
             workspace = workspace.with_embeddings(emb.clone());
         }
@@ -438,23 +629,6 @@ async fn main() -> anyhow::Result<()> {
         tracing::info!("Builder mode enabled");
     }
 
-    // Create secrets store if master key is configured (needed for MCP auth and WASM channels)
-    let secrets_store: Option<Arc<dyn SecretsStore + Send + Sync>> =
-        if let (Some(store), Some(master_key)) = (&store, config.secrets.master_key()) {
-            match SecretsCrypto::new(master_key.clone()) {
-                Ok(crypto) => Some(Arc::new(PostgresSecretsStore::new(
-                    store.pool(),
-                    Arc::new(crypto),
-                ))),
-                Err(e) => {
-                    tracing::warn!("Failed to initialize secrets crypto: {}", e);
-                    None
-                }
-            }
-        } else {
-            None
-        };
-
     let mcp_session_manager = Arc::new(McpSessionManager::new());
 
     // Create WASM tool runtime (sync, just builds the wasmtime engine)
@@ -475,7 +649,10 @@ async fn main() -> anyhow::Result<()> {
     // Both register into the shared ToolRegistry (RwLock-based) so concurrent writes are safe.
     let wasm_tools_future = async {
         if let Some(ref runtime) = wasm_tool_runtime {
-            let loader = WasmToolLoader::new(Arc::clone(runtime), Arc::clone(&tools));
+            let mut loader = WasmToolLoader::new(Arc::clone(runtime), Arc::clone(&tools));
+            if let Some(ref secrets) = secrets_store {
+                loader = loader.with_secrets_store(Arc::clone(secrets));
+            }
 
             // Load installed tools from ~/.ironclaw/tools/
             match loader.load_from_dir(&config.wasm.tools_dir).await {
@@ -515,8 +692,8 @@ async fn main() -> anyhow::Result<()> {
 
     let mcp_servers_future = async {
         if let Some(ref secrets) = secrets_store {
-            let servers_result = if let Some(ref s) = store {
-                load_mcp_servers_from_db(s, "default").await
+            let servers_result = if let Some(ref d) = db {
+                load_mcp_servers_from_db(d.as_ref(), "default").await
             } else {
                 ironclaw::tools::mcp::config::load_mcp_servers().await
             };
@@ -629,7 +806,7 @@ async fn main() -> anyhow::Result<()> {
             config.channels.wasm_channels_dir.clone(),
             config.tunnel.public_url.clone(),
             "default".to_string(),
-            store.clone(),
+            db.clone(),
         ));
         tools.register_extension_tools(Arc::clone(&manager));
         tracing::info!("Extension manager initialized with in-chat discovery tools");
@@ -681,6 +858,7 @@ async fn main() -> anyhow::Result<()> {
             claude_code_model: config.claude_code.model.clone(),
             claude_code_max_turns: config.claude_code.max_turns,
             claude_code_memory_limit_mb: config.claude_code.memory_limit_mb,
+            claude_code_allowed_tools: config.claude_code.allowed_tools.clone(),
         };
         let jm = Arc::new(ContainerJobManager::new(job_config, token_store.clone()));
 
@@ -691,7 +869,7 @@ async fn main() -> anyhow::Result<()> {
             token_store,
             job_event_tx: job_event_tx.clone(),
             prompt_queue: Arc::clone(&prompt_queue),
-            store: store.clone(),
+            store: db.clone(),
         };
 
         tokio::spawn(async move {
@@ -720,12 +898,14 @@ async fn main() -> anyhow::Result<()> {
 
     // Initialize channel manager
     let mut channels = ChannelManager::new();
+    let mut channel_names: Vec<String> = Vec::new();
 
     if let Some(repl) = repl_channel {
         channels.add(Box::new(repl));
         if cli.message.is_some() {
             tracing::info!("Single message mode");
         } else {
+            channel_names.push("repl".to_string());
             tracing::info!("REPL mode enabled");
         }
     }
@@ -797,13 +977,13 @@ async fn main() -> anyhow::Result<()> {
 
                                 // Inject owner_id for Telegram so the bot only responds
                                 // to the bound user account.
-                                if channel_name == "telegram" {
-                                    if let Some(owner_id) = config.channels.telegram_owner_id {
-                                        config_updates.insert(
-                                            "owner_id".to_string(),
-                                            serde_json::json!(owner_id),
-                                        );
-                                    }
+                                if channel_name == "telegram"
+                                    && let Some(owner_id) = config.channels.telegram_owner_id
+                                {
+                                    config_updates.insert(
+                                        "owner_id".to_string(),
+                                        serde_json::json!(owner_id),
+                                    );
                                 }
 
                                 if !config_updates.is_empty() {
@@ -861,6 +1041,7 @@ async fn main() -> anyhow::Result<()> {
                                 }
                             }
 
+                            channel_names.push(channel_name.clone());
                             channels.add(Box::new(SharedWasmChannel::new(channel_arc)));
                         }
 
@@ -894,23 +1075,24 @@ async fn main() -> anyhow::Result<()> {
     // Extract its routes for the unified server; the channel itself just
     // provides the mpsc stream.
     let mut webhook_server_addr: Option<std::net::SocketAddr> = None;
-    if !cli.cli_only {
-        if let Some(ref http_config) = config.channels.http {
-            let http_channel = HttpChannel::new(http_config.clone());
-            webhook_routes.push(http_channel.routes());
-            let (host, port) = http_channel.addr();
-            webhook_server_addr = Some(
-                format!("{}:{}", host, port)
-                    .parse()
-                    .expect("HttpConfig host:port must be a valid SocketAddr"),
-            );
-            channels.add(Box::new(http_channel));
-            tracing::info!(
-                "HTTP channel enabled on {}:{}",
-                http_config.host,
-                http_config.port
-            );
-        }
+    if !cli.cli_only
+        && let Some(ref http_config) = config.channels.http
+    {
+        let http_channel = HttpChannel::new(http_config.clone());
+        webhook_routes.push(http_channel.routes());
+        let (host, port) = http_channel.addr();
+        webhook_server_addr = Some(
+            format!("{}:{}", host, port)
+                .parse()
+                .expect("HttpConfig host:port must be a valid SocketAddr"),
+        );
+        channel_names.push("http".to_string());
+        channels.add(Box::new(http_channel));
+        tracing::info!(
+            "HTTP channel enabled on {}:{}",
+            http_config.host,
+            http_config.port
+        );
     }
 
     // Start the unified webhook server if any routes were registered.
@@ -928,13 +1110,15 @@ async fn main() -> anyhow::Result<()> {
     };
 
     // Create workspace for agent (shared with memory tools)
-    let workspace = store.as_ref().map(|s| {
-        let mut ws = Workspace::new("default", s.pool());
+    let workspace = if let Some(ref db_ref) = db {
+        let mut ws = Workspace::new_with_db("default", Arc::clone(db_ref));
         if let Some(ref emb) = embeddings {
             ws = ws.with_embeddings(emb.clone());
         }
-        Arc::new(ws)
-    });
+        Some(Arc::new(ws))
+    } else {
+        None
+    };
 
     // Seed workspace with core identity files on first boot
     if let Some(ref ws) = workspace {
@@ -965,17 +1149,21 @@ async fn main() -> anyhow::Result<()> {
     // Create context manager (shared between job tools and agent)
     let context_manager = Arc::new(ContextManager::new(config.agent.max_parallel_jobs));
 
+    // Create hook registry
+    let hooks = Arc::new(HookRegistry::new());
+
     // Create session manager (shared between agent and web gateway)
-    let session_manager = Arc::new(SessionManager::new());
+    let session_manager = Arc::new(SessionManager::new().with_hooks(hooks.clone()));
 
     // Register job tools (sandbox deps auto-injected when container_job_manager is available)
     tools.register_job_tools(
         Arc::clone(&context_manager),
         container_job_manager.clone(),
-        store.clone(),
+        db.clone(),
     );
 
     // Add web gateway channel if configured
+    let mut gateway_url: Option<String> = None;
     if let Some(ref gw_config) = config.channels.gateway {
         let mut gw = GatewayChannel::new(gw_config.clone());
         if let Some(ref ws) = workspace {
@@ -987,8 +1175,8 @@ async fn main() -> anyhow::Result<()> {
         if let Some(ref ext_mgr) = extension_manager {
             gw = gw.with_extension_manager(Arc::clone(ext_mgr));
         }
-        if let Some(ref s) = store {
-            gw = gw.with_store(Arc::clone(s));
+        if let Some(ref d) = db {
+            gw = gw.with_store(Arc::clone(d));
         }
         if let Some(ref jm) = container_job_manager {
             gw = gw.with_job_manager(Arc::clone(jm));
@@ -1008,29 +1196,39 @@ async fn main() -> anyhow::Result<()> {
             }
         }
 
+        gateway_url = Some(format!(
+            "http://{}:{}/?token={}",
+            gw_config.host,
+            gw_config.port,
+            gw.auth_token()
+        ));
+
         tracing::info!(
             "Web gateway enabled on {}:{}",
             gw_config.host,
             gw_config.port
         );
-        tracing::info!(
-            "Web UI: http://{}:{}/?token={}",
-            gw_config.host,
-            gw_config.port,
-            gw.auth_token()
-        );
+        tracing::info!("Web UI: http://{}:{}/", gw_config.host, gw_config.port);
 
+        channel_names.push("gateway".to_string());
         channels.add(Box::new(gw));
     }
 
+    // Capture boot screen info before moving Arcs into AgentDeps.
+    let boot_tool_count = tools.count();
+    let boot_llm_model = llm.model_name().to_string();
+    let boot_cheap_model = cheap_llm.as_ref().map(|c| c.model_name().to_string());
+
     // Create and run the agent
     let deps = AgentDeps {
-        store,
+        store: db,
         llm,
+        cheap_llm,
         safety,
         tools,
         workspace,
         extension_manager,
+        hooks,
     };
     let agent = Agent::new(
         config.agent.clone(),
@@ -1044,6 +1242,38 @@ async fn main() -> anyhow::Result<()> {
 
     tracing::info!("Agent initialized, starting main loop...");
 
+    // Print boot screen for interactive CLI mode (not single-message mode).
+    if config.channels.cli.enabled && cli.message.is_none() {
+        let boot_info = ironclaw::boot_screen::BootInfo {
+            version: env!("CARGO_PKG_VERSION").to_string(),
+            agent_name: config.agent.name.clone(),
+            llm_backend: config.llm.backend.to_string(),
+            llm_model: boot_llm_model,
+            cheap_model: boot_cheap_model,
+            db_backend: if cli.no_db {
+                "none".to_string()
+            } else {
+                config.database.backend.to_string()
+            },
+            db_connected: !cli.no_db,
+            tool_count: boot_tool_count,
+            gateway_url,
+            embeddings_enabled: config.embeddings.enabled,
+            embeddings_provider: if config.embeddings.enabled {
+                Some(config.embeddings.provider.clone())
+            } else {
+                None
+            },
+            heartbeat_enabled: config.heartbeat.enabled,
+            heartbeat_interval_secs: config.heartbeat.interval_secs,
+            sandbox_enabled: config.sandbox.enabled,
+            claude_code_enabled: config.claude_code.enabled,
+            routines_enabled: config.routines.enabled,
+            channels: channel_names,
+        };
+        ironclaw::boot_screen::print_boot_screen(&boot_info);
+    }
+
     // Run the agent (blocks until shutdown)
     agent.run().await?;
 
@@ -1059,27 +1289,28 @@ async fn main() -> anyhow::Result<()> {
 /// Check if onboarding is needed and return the reason.
 ///
 /// Returns `Some(reason)` if onboarding should be triggered, `None` otherwise.
-async fn check_onboard_needed() -> Option<&'static str> {
-    let bootstrap = ironclaw::bootstrap::BootstrapConfig::load();
+/// Called after `load_ironclaw_env()`, so DATABASE_URL from `~/.ironclaw/.env`
+/// is already in the environment.
+#[cfg(any(feature = "postgres", feature = "libsql"))]
+fn check_onboard_needed() -> Option<&'static str> {
+    let has_db = std::env::var("DATABASE_URL").is_ok()
+        || std::env::var("LIBSQL_PATH").is_ok()
+        || ironclaw::config::default_libsql_path().exists();
 
-    // Database not configured (and not in env)
-    if bootstrap.database_url.is_none() && std::env::var("DATABASE_URL").is_err() {
+    if !has_db {
         return Some("Database not configured");
     }
 
-    // Secrets not configured (and not in env)
-    if bootstrap.secrets_master_key_source == ironclaw::settings::KeySource::None
-        && std::env::var("SECRETS_MASTER_KEY").is_err()
-        && !ironclaw::secrets::keychain::has_master_key().await
-    {
-        // Only require secrets setup if user hasn't explicitly disabled it
-        // For now, we don't require it for first run
-    }
-
-    // First run (onboarding never completed and no session)
-    let session_path = ironclaw::llm::session::default_session_path();
-    if !bootstrap.onboard_completed && !session_path.exists() {
-        return Some("First run");
+    // First run (onboarding never completed and no session).
+    // Reads NEARAI_API_KEY env var directly because this function runs
+    // before Config is loaded -- Config::from_env() may fail without a
+    // database URL, which is what triggers onboarding in the first place.
+    if std::env::var("NEARAI_API_KEY").is_err() {
+        let settings = ironclaw::settings::Settings::load();
+        let session_path = ironclaw::llm::session::default_session_path();
+        if !settings.onboard_completed && !session_path.exists() {
+            return Some("First run");
+        }
     }
 
     None

src/orchestrator/api.rs

@@ -15,7 +15,7 @@ use tokio::sync::{Mutex, broadcast};
 use uuid::Uuid;
 
 use crate::channels::web::types::SseEvent;
-use crate::history::Store;
+use crate::db::Database;
 use crate::llm::{CompletionRequest, LlmProvider, ToolCompletionRequest};
 use crate::orchestrator::auth::{TokenStore, worker_auth_middleware};
 use crate::orchestrator::job_manager::ContainerJobManager;
@@ -43,7 +43,7 @@ pub struct OrchestratorState {
     /// Buffered follow-up prompts for sandbox jobs, keyed by job_id.
     pub prompt_queue: Arc<Mutex<HashMap<Uuid, VecDeque<PendingPrompt>>>>,
     /// Database handle for persisting job events.
-    pub store: Option<Arc<Store>>,
+    pub store: Option<Arc<dyn Database>>,
 }
 
 /// The orchestrator's internal API server.
@@ -202,7 +202,7 @@ async fn report_complete(
     State(state): State<OrchestratorState>,
     Path(job_id): Path<Uuid>,
     Json(report): Json<CompletionReport>,
-) -> Result<StatusCode, StatusCode> {
+) -> Result<Json<serde_json::Value>, StatusCode> {
     if report.success {
         tracing::info!(
             job_id = %job_id,
@@ -223,7 +223,7 @@ async fn report_complete(
     };
     let _ = state.job_manager.complete_job(job_id, result).await;
 
-    Ok(StatusCode::OK)
+    Ok(Json(serde_json::json!({"status": "ok"})))
 }
 
 // -- Sandbox job event handlers --
@@ -339,16 +339,16 @@ async fn get_prompt_handler(
     Path(job_id): Path<Uuid>,
 ) -> Result<(StatusCode, Json<serde_json::Value>), StatusCode> {
     let mut queue = state.prompt_queue.lock().await;
-    if let Some(prompts) = queue.get_mut(&job_id) {
-        if let Some(prompt) = prompts.pop_front() {
-            return Ok((
-                StatusCode::OK,
-                Json(serde_json::json!({
-                    "content": prompt.content,
-                    "done": prompt.done,
-                })),
-            ));
-        }
+    if let Some(prompts) = queue.get_mut(&job_id)
+        && let Some(prompt) = prompts.pop_front()
+    {
+        return Ok((
+            StatusCode::OK,
+            Json(serde_json::json!({
+                "content": prompt.content,
+                "done": prompt.done,
+            })),
+        ));
     }
 
     // Return 204 with an empty body. The Json wrapper requires some value

src/orchestrator/auth.rs

@@ -14,6 +14,7 @@ use axum::http::StatusCode;
 use axum::middleware::Next;
 use axum::response::Response;
 use rand::Rng;
+use subtle::ConstantTimeEq;
 use tokio::sync::RwLock;
 use uuid::Uuid;
 
@@ -38,13 +39,13 @@ impl TokenStore {
         token
     }
 
-    /// Validate a token for a specific job.
+    /// Validate a token for a specific job (constant-time comparison).
     pub async fn validate(&self, job_id: Uuid, token: &str) -> bool {
         self.tokens
             .read()
             .await
             .get(&job_id)
-            .map(|stored| stored == token)
+            .map(|stored| stored.as_bytes().ct_eq(token.as_bytes()).into())
             .unwrap_or(false)
     }
 

src/orchestrator/job_manager.rs

@@ -58,6 +58,8 @@ pub struct ContainerJobConfig {
     pub claude_code_max_turns: u32,
     /// Memory limit in MB for Claude Code containers (heavier than workers).
     pub claude_code_memory_limit_mb: u64,
+    /// Allowed tool patterns for Claude Code (passed as CLAUDE_CODE_ALLOWED_TOOLS env var).
+    pub claude_code_allowed_tools: Vec<String>,
 }
 
 impl Default for ContainerJobConfig {
@@ -71,6 +73,7 @@ impl Default for ContainerJobConfig {
             claude_code_model: "sonnet".to_string(),
             claude_code_max_turns: 50,
             claude_code_memory_limit_mb: 4096,
+            claude_code_allowed_tools: crate::config::ClaudeCodeConfig::default().allowed_tools,
         }
     }
 }
@@ -161,6 +164,29 @@ impl ContainerJobManager {
         };
         self.containers.write().await.insert(job_id, handle);
 
+        // Run the actual container creation. On any failure, revoke the token
+        // and remove the handle so we don't leak resources.
+        match self
+            .create_job_inner(job_id, &token, project_dir, mode)
+            .await
+        {
+            Ok(()) => Ok(token),
+            Err(e) => {
+                self.token_store.revoke(job_id).await;
+                self.containers.write().await.remove(&job_id);
+                Err(e)
+            }
+        }
+    }
+
+    /// Inner implementation of container creation (separated for cleanup).
+    async fn create_job_inner(
+        &self,
+        job_id: Uuid,
+        token: &str,
+        project_dir: Option<PathBuf>,
+        mode: JobMode,
+    ) -> Result<(), OrchestratorError> {
         // Connect to Docker
         let docker = connect_docker()
             .await
@@ -203,27 +229,34 @@ impl ContainerJobManager {
                 .unwrap_or_else(|| PathBuf::from("."))
                 .join(".ironclaw")
                 .join("projects");
-            if let Ok(canonical_base) = projects_base.canonicalize() {
-                if !canonical.starts_with(&canonical_base) {
-                    return Err(OrchestratorError::ContainerCreationFailed {
-                        job_id,
-                        reason: format!(
-                            "project directory {} is outside allowed base {}",
-                            canonical.display(),
-                            canonical_base.display()
-                        ),
-                    });
-                }
+            if let Ok(canonical_base) = projects_base.canonicalize()
+                && !canonical.starts_with(&canonical_base)
+            {
+                return Err(OrchestratorError::ContainerCreationFailed {
+                    job_id,
+                    reason: format!(
+                        "project directory {} is outside allowed base {}",
+                        canonical.display(),
+                        canonical_base.display()
+                    ),
+                });
             }
             binds.push(format!("{}:/workspace:rw", canonical.display()));
             env_vec.push("IRONCLAW_WORKSPACE=/workspace".to_string());
         }
 
-        // Claude Code mode: mount host ~/.claude read-only for auth
+        // Claude Code mode: mount host ~/.claude read-only for auth,
+        // and pass the tool allowlist so the bridge can write settings.json.
         if mode == JobMode::ClaudeCode {
             if let Some(ref claude_dir) = self.config.claude_config_dir {
                 binds.push(format!("{}:/home/sandbox/.claude:ro", claude_dir.display()));
             }
+            if !self.config.claude_code_allowed_tools.is_empty() {
+                env_vec.push(format!(
+                    "CLAUDE_CODE_ALLOWED_TOOLS={}",
+                    self.config.claude_code_allowed_tools.join(",")
+                ));
+            }
         }
 
         // Memory limit: Claude Code gets more memory
@@ -243,11 +276,7 @@ impl ContainerJobManager {
             network_mode: Some("bridge".to_string()),
             extra_hosts: Some(vec!["host.docker.internal:host-gateway".to_string()]),
             cap_drop: Some(vec!["ALL".to_string()]),
-            cap_add: Some(vec![
-                "CHOWN".to_string(),
-                "SETUID".to_string(),
-                "SETGID".to_string(),
-            ]),
+            cap_add: Some(vec!["CHOWN".to_string()]),
             security_opt: Some(vec!["no-new-privileges:true".to_string()]),
             tmpfs: Some(
                 [("/tmp".to_string(), "size=512M".to_string())]
@@ -328,7 +357,7 @@ impl ContainerJobManager {
             "Created and started worker container"
         );
 
-        Ok(token)
+        Ok(())
     }
 
     /// Stop a running container job.
@@ -355,15 +384,18 @@ impl ContainerJobManager {
             })?;
 
         // Stop the container (10 second grace period)
-        let _ = docker
+        if let Err(e) = docker
             .stop_container(
                 &container_id,
                 Some(bollard::container::StopContainerOptions { t: 10 }),
             )
-            .await;
+            .await
+        {
+            tracing::warn!(job_id = %job_id, error = %e, "Failed to stop container (may already be stopped)");
+        }
 
         // Remove the container
-        let _ = docker
+        if let Err(e) = docker
             .remove_container(
                 &container_id,
                 Some(bollard::container::RemoveContainerOptions {
@@ -371,7 +403,10 @@ impl ContainerJobManager {
                     ..Default::default()
                 }),
             )
-            .await;
+            .await
+        {
+            tracing::warn!(job_id = %job_id, error = %e, "Failed to remove container (may require manual cleanup)");
+        }
 
         // Update state
         if let Some(handle) = self.containers.write().await.get_mut(&job_id) {
@@ -407,16 +442,21 @@ impl ContainerJobManager {
             let containers = self.containers.read().await;
             containers.get(&job_id).map(|h| h.container_id.clone())
         };
-        if let Some(cid) = container_id {
-            if !cid.is_empty() {
-                if let Ok(docker) = connect_docker().await {
-                    let _ = docker
+        if let Some(cid) = container_id
+            && !cid.is_empty()
+        {
+            match connect_docker().await {
+                Ok(docker) => {
+                    if let Err(e) = docker
                         .stop_container(
                             &cid,
                             Some(bollard::container::StopContainerOptions { t: 5 }),
                         )
-                        .await;
-                    let _ = docker
+                        .await
+                    {
+                        tracing::warn!(job_id = %job_id, error = %e, "Failed to stop completed container");
+                    }
+                    if let Err(e) = docker
                         .remove_container(
                             &cid,
                             Some(bollard::container::RemoveContainerOptions {
@@ -424,7 +464,13 @@ impl ContainerJobManager {
                                 ..Default::default()
                             }),
                         )
-                        .await;
+                        .await
+                    {
+                        tracing::warn!(job_id = %job_id, error = %e, "Failed to remove completed container");
+                    }
+                }
+                Err(e) => {
+                    tracing::warn!(job_id = %job_id, error = %e, "Failed to connect to Docker for container cleanup");
                 }
             }
         }

src/pairing/store.rs

@@ -320,19 +320,27 @@ impl PairingStore {
     fn record_failed_approve(&self, channel: &str) -> Result<(), PairingStoreError> {
         let path = approve_attempts_path(&self.base_dir, channel)?;
         fs::create_dir_all(path.parent().unwrap())?;
+
+        // Open (or create) and lock before reading so concurrent callers
+        // don't clobber each other's writes.
         let file = fs::OpenOptions::new()
             .read(true)
             .write(true)
             .create(true)
-            .truncate(true)
+            .truncate(false)
             .open(&path)?;
         file.lock_exclusive()?;
-        let content = fs::read_to_string(&path).unwrap_or_default();
-        let mut data: ApproveAttemptsFile = serde_json::from_str(&content).unwrap_or_default();
+
+        let mut data: ApproveAttemptsFile = fs::read_to_string(&path)
+            .ok()
+            .and_then(|c| serde_json::from_str(&c).ok())
+            .unwrap_or_default();
+
         let now = now_secs();
         data.failed_at.push(now);
         let cutoff = now.saturating_sub(PAIRING_APPROVE_RATE_WINDOW_SECS);
         data.failed_at.retain(|&t| t >= cutoff);
+
         let json = serde_json::to_string_pretty(&data)?;
         fs::write(&path, json)?;
         fs4::FileExt::unlock(&file)?;

src/safety/leak_detector.rs

@@ -147,10 +147,10 @@ impl LeakDetector {
         // Build prefix matcher for patterns that start with a known prefix
         let mut prefixes = Vec::new();
         for (idx, pattern) in patterns.iter().enumerate() {
-            if let Some(prefix) = extract_literal_prefix(pattern.regex.as_str()) {
-                if prefix.len() >= 3 {
-                    prefixes.push((prefix, idx));
-                }
+            if let Some(prefix) = extract_literal_prefix(pattern.regex.as_str())
+                && prefix.len() >= 3
+            {
+                prefixes.push((prefix, idx));
             }
         }
 
@@ -306,12 +306,11 @@ impl LeakDetector {
                 })?;
         }
 
-        // Scan body if present and valid UTF-8
+        // Scan body if present. Use lossy UTF-8 conversion so a leading
+        // non-UTF8 byte can't be used to skip scanning entirely.
         if let Some(body_bytes) = body {
-            if let Ok(body_str) = std::str::from_utf8(body_bytes) {
-                self.scan_and_clean(body_str)?;
-            }
-            // Binary bodies are not scanned (could add hex pattern detection later)
+            let body_str = String::from_utf8_lossy(body_bytes);
+            self.scan_and_clean(&body_str)?;
         }
 
         Ok(())
@@ -705,4 +704,17 @@ mod tests {
         let result = detector.scan_http_request("https://api.example.com/webhook", &[], Some(body));
         assert!(result.is_err());
     }
+
+    #[test]
+    fn test_scan_http_request_blocks_secret_in_binary_body() {
+        let detector = LeakDetector::new();
+
+        // Attacker prepends a non-UTF8 byte to bypass strict from_utf8 check.
+        // The lossy conversion should still detect the secret.
+        let mut body = vec![0xFF]; // invalid UTF-8 leading byte
+        body.extend_from_slice(b"sk-proj-test1234567890abcdefghij");
+
+        let result = detector.scan_http_request("https://api.example.com/exfil", &[], Some(&body));
+        assert!(result.is_err(), "binary body should still be scanned");
+    }
 }

src/safety/mod.rs

@@ -98,15 +98,15 @@ impl SafetyLayer {
                 was_modified: true,
             };
         }
-        if violations
+        let force_sanitize = violations
             .iter()
-            .any(|rule| rule.action == crate::safety::PolicyAction::Sanitize)
-        {
+            .any(|rule| rule.action == crate::safety::PolicyAction::Sanitize);
+        if force_sanitize {
             was_modified = true;
         }
 
-        // Run sanitization if enabled
-        if self.config.injection_check_enabled {
+        // Run sanitization once: if injection_check is enabled OR policy requires it
+        if self.config.injection_check_enabled || force_sanitize {
             let mut sanitized = self.sanitizer.sanitize(&content);
             sanitized.was_modified = sanitized.was_modified || was_modified;
             sanitized
@@ -190,4 +190,20 @@ mod tests {
         assert!(wrapped.contains("sanitized=\"true\""));
         assert!(wrapped.contains("Hello <world>"));
     }
+
+    #[test]
+    fn test_sanitize_action_forces_sanitization_when_injection_check_disabled() {
+        let config = SafetyConfig {
+            max_output_length: 100_000,
+            injection_check_enabled: false,
+        };
+        let safety = SafetyLayer::new(&config);
+
+        // Content with an injection-like pattern that a policy might flag
+        let output = safety.sanitize_tool_output("test", "normal text");
+        // With injection_check disabled and no policy violations, content
+        // should pass through unmodified
+        assert_eq!(output.content, "normal text");
+        assert!(!output.was_modified);
+    }
 }

src/sandbox/container.rs

@@ -279,11 +279,7 @@ impl ContainerRunner {
             network_mode: Some("bridge".to_string()),
             // Security: drop all capabilities and add back only what's needed
             cap_drop: Some(vec!["ALL".to_string()]),
-            cap_add: Some(vec![
-                "CHOWN".to_string(),
-                "SETUID".to_string(),
-                "SETGID".to_string(),
-            ]),
+            cap_add: Some(vec!["CHOWN".to_string()]),
             // Prevent privilege escalation
             security_opt: Some(vec!["no-new-privileges:true".to_string()]),
             // Read-only root filesystem (workspace is still writable if policy allows)
@@ -498,10 +494,10 @@ impl ContainerRunner {
 /// 3. `~/.docker/run/docker.sock` (Docker Desktop on macOS)
 pub async fn connect_docker() -> Result<Docker> {
     // First try bollard defaults (checks DOCKER_HOST, then /var/run/docker.sock)
-    if let Ok(docker) = Docker::connect_with_local_defaults() {
-        if docker.ping().await.is_ok() {
-            return Ok(docker);
-        }
+    if let Ok(docker) = Docker::connect_with_local_defaults()
+        && docker.ping().await.is_ok()
+    {
+        return Ok(docker);
     }
 
     // Try Docker Desktop socket (macOS)
@@ -511,10 +507,9 @@ pub async fn connect_docker() -> Result<Docker> {
             let sock_str = desktop_sock.to_string_lossy();
             if let Ok(docker) =
                 Docker::connect_with_socket(&sock_str, 120, bollard::API_DEFAULT_VERSION)
+                && docker.ping().await.is_ok()
             {
-                if docker.ping().await.is_ok() {
-                    return Ok(docker);
-                }
+                return Ok(docker);
             }
         }
     }

src/sandbox/proxy/http.rs

@@ -259,11 +259,11 @@ async fn handle_connect(
 
     let decision = state.decider.decide(&network_req).await;
 
-    if !decision.is_allowed() {
-        if let NetworkDecision::Deny { reason } = decision {
-            tracing::info!("Proxy: blocked CONNECT {} - {}", host, reason);
-            return error_response(StatusCode::FORBIDDEN, reason);
-        }
+    if !decision.is_allowed()
+        && let NetworkDecision::Deny { reason } = decision
+    {
+        tracing::info!("Proxy: blocked CONNECT {} - {}", host, reason);
+        return error_response(StatusCode::FORBIDDEN, reason);
     }
 
     tracing::debug!("Proxy: allowing CONNECT to {}", host);
@@ -294,10 +294,10 @@ async fn forward_request(
 
     // Copy headers (except hop-by-hop headers)
     for (name, value) in req.headers() {
-        if !is_hop_by_hop_header(name.as_str()) {
-            if let Ok(v) = value.to_str() {
-                builder = builder.header(name.as_str(), v);
-            }
+        if !is_hop_by_hop_header(name.as_str())
+            && let Ok(v) = value.to_str()
+        {
+            builder = builder.header(name.as_str(), v);
         }
     }
 

src/sandbox/proxy/policy.rs

@@ -109,12 +109,11 @@ impl NetworkPolicyDecider for DefaultPolicyDecider {
     async fn decide(&self, request: &NetworkRequest) -> NetworkDecision {
         // First check if the domain is allowed
         let validation = self.allowlist.is_allowed(&request.host);
-        if !validation.is_allowed() {
-            if let crate::sandbox::proxy::allowlist::DomainValidationResult::Denied(reason) =
+        if !validation.is_allowed()
+            && let crate::sandbox::proxy::allowlist::DomainValidationResult::Denied(reason) =
                 validation
-            {
-                return NetworkDecision::Deny { reason };
-            }
+        {
+            return NetworkDecision::Deny { reason };
         }
 
         // Check if we need to inject credentials

src/secrets/keychain.rs

@@ -261,7 +261,7 @@ pub use platform::{delete_master_key, get_master_key, has_master_key, store_mast
 
 /// Parse a hex string to bytes.
 fn hex_to_bytes(hex: &str) -> Result<Vec<u8>, SecretError> {
-    if hex.len() % 2 != 0 {
+    if !hex.len().is_multiple_of(2) {
         return Err(SecretError::KeychainError(
             "Invalid hex string length".to_string(),
         ));

src/secrets/mod.rs

@@ -64,7 +64,11 @@ mod store;
 mod types;
 
 pub use crypto::SecretsCrypto;
-pub use store::{PostgresSecretsStore, SecretsStore};
+#[cfg(feature = "libsql")]
+pub use store::LibSqlSecretsStore;
+#[cfg(feature = "postgres")]
+pub use store::PostgresSecretsStore;
+pub use store::SecretsStore;
 pub use types::{
     CreateSecretParams, CredentialLocation, CredentialMapping, DecryptedSecret, Secret,
     SecretError, SecretRef,

src/secrets/store.rs

@@ -10,6 +10,7 @@ use std::sync::Arc;
 
 use async_trait::async_trait;
 use chrono::Utc;
+#[cfg(feature = "postgres")]
 use deadpool_postgres::Pool;
 use secrecy::ExposeSecret;
 use uuid::Uuid;
@@ -61,11 +62,13 @@ pub trait SecretsStore: Send + Sync {
 }
 
 /// PostgreSQL implementation of SecretsStore.
+#[cfg(feature = "postgres")]
 pub struct PostgresSecretsStore {
     pool: Pool,
     crypto: Arc<SecretsCrypto>,
 }
 
+#[cfg(feature = "postgres")]
 impl PostgresSecretsStore {
     /// Create a new store with the given database pool and crypto instance.
     pub fn new(pool: Pool, crypto: Arc<SecretsCrypto>) -> Self {
@@ -73,6 +76,7 @@ impl PostgresSecretsStore {
     }
 }
 
+#[cfg(feature = "postgres")]
 #[async_trait]
 impl SecretsStore for PostgresSecretsStore {
     async fn create(
@@ -149,10 +153,10 @@ impl SecretsStore for PostgresSecretsStore {
                 let secret = row_to_secret(&r);
 
                 // Check expiration
-                if let Some(expires_at) = secret.expires_at {
-                    if expires_at < Utc::now() {
-                        return Err(SecretError::Expired);
-                    }
+                if let Some(expires_at) = secret.expires_at
+                    && expires_at < Utc::now()
+                {
+                    return Err(SecretError::Expired);
                 }
 
                 Ok(secret)
@@ -272,10 +276,10 @@ impl SecretsStore for PostgresSecretsStore {
             }
 
             // Simple glob: * matches any suffix
-            if let Some(prefix) = pattern.strip_suffix('*') {
-                if secret_name.starts_with(prefix) {
-                    return Ok(true);
-                }
+            if let Some(prefix) = pattern.strip_suffix('*')
+                && secret_name.starts_with(prefix)
+            {
+                return Ok(true);
             }
         }
 
@@ -283,6 +287,7 @@ impl SecretsStore for PostgresSecretsStore {
     }
 }
 
+#[cfg(feature = "postgres")]
 fn row_to_secret(row: &tokio_postgres::Row) -> Secret {
     Secret {
         id: row.get("id"),
@@ -299,6 +304,332 @@ fn row_to_secret(row: &tokio_postgres::Row) -> Secret {
     }
 }
 
+// ==================== libSQL implementation ====================
+
+/// libSQL/Turso implementation of SecretsStore.
+///
+/// Holds an `Arc<Database>` handle and creates a fresh connection per operation,
+/// matching the connection-per-request pattern used by the main `LibSqlBackend`.
+#[cfg(feature = "libsql")]
+pub struct LibSqlSecretsStore {
+    db: Arc<libsql::Database>,
+    crypto: Arc<SecretsCrypto>,
+}
+
+#[cfg(feature = "libsql")]
+impl LibSqlSecretsStore {
+    /// Create a new store with the given shared libsql database handle and crypto instance.
+    pub fn new(db: Arc<libsql::Database>, crypto: Arc<SecretsCrypto>) -> Self {
+        Self { db, crypto }
+    }
+
+    fn connect(&self) -> Result<libsql::Connection, SecretError> {
+        self.db
+            .connect()
+            .map_err(|e| SecretError::Database(format!("Connection failed: {}", e)))
+    }
+}
+
+#[cfg(feature = "libsql")]
+#[async_trait]
+impl SecretsStore for LibSqlSecretsStore {
+    async fn create(
+        &self,
+        user_id: &str,
+        params: CreateSecretParams,
+    ) -> Result<Secret, SecretError> {
+        let plaintext = params.value.expose_secret().as_bytes();
+        let (encrypted_value, key_salt) = self.crypto.encrypt(plaintext)?;
+
+        let id = Uuid::new_v4();
+        let now = Utc::now();
+        let now_str = now.to_rfc3339_opts(chrono::SecondsFormat::Millis, true);
+        let expires_at_str = params
+            .expires_at
+            .map(|dt| dt.to_rfc3339_opts(chrono::SecondsFormat::Millis, true));
+
+        // Start transaction for atomic upsert + read-back
+        let conn = self.connect()?;
+        let tx = conn
+            .transaction()
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        tx.execute(
+                r#"
+                INSERT INTO secrets (id, user_id, name, encrypted_value, key_salt, provider, expires_at, created_at, updated_at)
+                VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?8)
+                ON CONFLICT (user_id, name) DO UPDATE SET
+                    encrypted_value = excluded.encrypted_value,
+                    key_salt = excluded.key_salt,
+                    provider = excluded.provider,
+                    expires_at = excluded.expires_at,
+                    updated_at = ?8
+                "#,
+                libsql::params![
+                    id.to_string(),
+                    user_id,
+                    params.name.as_str(),
+                    libsql::Value::Blob(encrypted_value.clone()),
+                    libsql::Value::Blob(key_salt.clone()),
+                    libsql_opt_text(params.provider.as_deref()),
+                    libsql_opt_text(expires_at_str.as_deref()),
+                    now_str.as_str(),
+                ],
+            )
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        // Read back the row (may have been upserted)
+        let mut rows = tx
+            .query(
+                r#"
+                SELECT id, user_id, name, encrypted_value, key_salt, provider, expires_at,
+                       last_used_at, usage_count, created_at, updated_at
+                FROM secrets
+                WHERE user_id = ?1 AND name = ?2
+                "#,
+                libsql::params![user_id, params.name.as_str()],
+            )
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        let row = rows
+            .next()
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?
+            .ok_or_else(|| SecretError::Database("Insert succeeded but row not found".into()))?;
+
+        let secret = libsql_row_to_secret(&row)?;
+
+        tx.commit()
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        Ok(secret)
+    }
+
+    async fn get(&self, user_id: &str, name: &str) -> Result<Secret, SecretError> {
+        let conn = self.connect()?;
+        let mut rows = conn
+            .query(
+                r#"
+                SELECT id, user_id, name, encrypted_value, key_salt, provider, expires_at,
+                       last_used_at, usage_count, created_at, updated_at
+                FROM secrets
+                WHERE user_id = ?1 AND name = ?2
+                "#,
+                libsql::params![user_id, name],
+            )
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        match rows
+            .next()
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?
+        {
+            Some(row) => {
+                let secret = libsql_row_to_secret(&row)?;
+
+                if let Some(expires_at) = secret.expires_at
+                    && expires_at < Utc::now()
+                {
+                    return Err(SecretError::Expired);
+                }
+
+                Ok(secret)
+            }
+            None => Err(SecretError::NotFound(name.to_string())),
+        }
+    }
+
+    async fn get_decrypted(
+        &self,
+        user_id: &str,
+        name: &str,
+    ) -> Result<DecryptedSecret, SecretError> {
+        let secret = self.get(user_id, name).await?;
+        self.crypto
+            .decrypt(&secret.encrypted_value, &secret.key_salt)
+    }
+
+    async fn exists(&self, user_id: &str, name: &str) -> Result<bool, SecretError> {
+        let conn = self.connect()?;
+        let mut rows = conn
+            .query(
+                "SELECT 1 FROM secrets WHERE user_id = ?1 AND name = ?2",
+                libsql::params![user_id, name],
+            )
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        Ok(rows
+            .next()
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?
+            .is_some())
+    }
+
+    async fn list(&self, user_id: &str) -> Result<Vec<SecretRef>, SecretError> {
+        let conn = self.connect()?;
+        let mut rows = conn
+            .query(
+                "SELECT name, provider FROM secrets WHERE user_id = ?1 ORDER BY name",
+                libsql::params![user_id],
+            )
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        let mut refs = Vec::new();
+        while let Some(row) = rows
+            .next()
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?
+        {
+            refs.push(SecretRef {
+                name: row.get::<String>(0).unwrap_or_default(),
+                provider: row.get::<String>(1).ok(),
+            });
+        }
+        Ok(refs)
+    }
+
+    async fn delete(&self, user_id: &str, name: &str) -> Result<bool, SecretError> {
+        let conn = self.connect()?;
+        let affected = conn
+            .execute(
+                "DELETE FROM secrets WHERE user_id = ?1 AND name = ?2",
+                libsql::params![user_id, name],
+            )
+            .await
+            .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        Ok(affected > 0)
+    }
+
+    async fn record_usage(&self, secret_id: Uuid) -> Result<(), SecretError> {
+        let now = Utc::now().to_rfc3339_opts(chrono::SecondsFormat::Millis, true);
+        let conn = self.connect()?;
+
+        conn.execute(
+            r#"
+                UPDATE secrets
+                SET last_used_at = ?1, usage_count = usage_count + 1
+                WHERE id = ?2
+                "#,
+            libsql::params![now.as_str(), secret_id.to_string()],
+        )
+        .await
+        .map_err(|e| SecretError::Database(e.to_string()))?;
+
+        Ok(())
+    }
+
+    async fn is_accessible(
+        &self,
+        user_id: &str,
+        secret_name: &str,
+        allowed_secrets: &[String],
+    ) -> Result<bool, SecretError> {
+        if !self.exists(user_id, secret_name).await? {
+            return Ok(false);
+        }
+
+        for pattern in allowed_secrets {
+            if pattern == secret_name {
+                return Ok(true);
+            }
+
+            if let Some(prefix) = pattern.strip_suffix('*')
+                && secret_name.starts_with(prefix)
+            {
+                return Ok(true);
+            }
+        }
+
+        Ok(false)
+    }
+}
+
+#[cfg(feature = "libsql")]
+fn libsql_opt_text(s: Option<&str>) -> libsql::Value {
+    match s {
+        Some(s) => libsql::Value::Text(s.to_string()),
+        None => libsql::Value::Null,
+    }
+}
+
+#[cfg(feature = "libsql")]
+fn libsql_parse_timestamp(s: &str) -> Result<chrono::DateTime<Utc>, SecretError> {
+    if let Ok(dt) = chrono::DateTime::parse_from_rfc3339(s) {
+        return Ok(dt.with_timezone(&Utc));
+    }
+    if let Ok(ndt) = chrono::NaiveDateTime::parse_from_str(s, "%Y-%m-%d %H:%M:%S%.f") {
+        return Ok(ndt.and_utc());
+    }
+    if let Ok(ndt) = chrono::NaiveDateTime::parse_from_str(s, "%Y-%m-%d %H:%M:%S") {
+        return Ok(ndt.and_utc());
+    }
+    Err(SecretError::Database(format!(
+        "unparseable timestamp: {:?}",
+        s
+    )))
+}
+
+#[cfg(feature = "libsql")]
+fn libsql_row_to_secret(row: &libsql::Row) -> Result<Secret, SecretError> {
+    let id_str: String = row
+        .get(0)
+        .map_err(|e| SecretError::Database(e.to_string()))?;
+    let user_id: String = row
+        .get(1)
+        .map_err(|e| SecretError::Database(e.to_string()))?;
+    let name: String = row
+        .get(2)
+        .map_err(|e| SecretError::Database(e.to_string()))?;
+    let encrypted_value: Vec<u8> = row
+        .get(3)
+        .map_err(|e| SecretError::Database(e.to_string()))?;
+    let key_salt: Vec<u8> = row
+        .get(4)
+        .map_err(|e| SecretError::Database(e.to_string()))?;
+    let provider: Option<String> = row.get::<String>(5).ok().filter(|s| !s.is_empty());
+    let expires_at = row
+        .get::<String>(6)
+        .ok()
+        .filter(|s| !s.is_empty())
+        .and_then(|s| libsql_parse_timestamp(&s).ok());
+    let last_used_at = row
+        .get::<String>(7)
+        .ok()
+        .filter(|s| !s.is_empty())
+        .and_then(|s| libsql_parse_timestamp(&s).ok());
+    let usage_count: i64 = row.get::<i64>(8).unwrap_or(0);
+    let created_at_str: String = row
+        .get(9)
+        .map_err(|e| SecretError::Database(e.to_string()))?;
+    let updated_at_str: String = row
+        .get(10)
+        .map_err(|e| SecretError::Database(e.to_string()))?;
+
+    Ok(Secret {
+        id: id_str
+            .parse()
+            .map_err(|e: uuid::Error| SecretError::Database(e.to_string()))?,
+        user_id,
+        name,
+        encrypted_value,
+        key_salt,
+        provider,
+        expires_at,
+        last_used_at,
+        usage_count,
+        created_at: libsql_parse_timestamp(&created_at_str)?,
+        updated_at: libsql_parse_timestamp(&updated_at_str)?,
+    })
+}
+
 /// In-memory implementation for testing.
 #[cfg(test)]
 pub mod testing {
@@ -364,12 +695,21 @@ pub mod testing {
         }
 
         async fn get(&self, user_id: &str, name: &str) -> Result<Secret, SecretError> {
-            self.secrets
+            let secret = self
+                .secrets
                 .read()
                 .await
                 .get(&(user_id.to_string(), name.to_string()))
                 .cloned()
-                .ok_or_else(|| SecretError::NotFound(name.to_string()))
+                .ok_or_else(|| SecretError::NotFound(name.to_string()))?;
+
+            if let Some(expires_at) = secret.expires_at
+                && expires_at < Utc::now()
+            {
+                return Err(SecretError::Expired);
+            }
+
+            Ok(secret)
         }
 
         async fn get_decrypted(
@@ -430,10 +770,10 @@ pub mod testing {
                 if pattern == secret_name {
                     return Ok(true);
                 }
-                if let Some(prefix) = pattern.strip_suffix('*') {
-                    if secret_name.starts_with(prefix) {
-                        return Ok(true);
-                    }
+                if let Some(prefix) = pattern.strip_suffix('*')
+                    && secret_name.starts_with(prefix)
+                {
+                    return Ok(true);
                 }
             }
             Ok(false)
@@ -558,6 +898,34 @@ mod tests {
         );
     }
 
+    #[tokio::test]
+    async fn test_expired_secret_returns_error() {
+        let store = test_store();
+        let expires_at = chrono::Utc::now() - chrono::Duration::hours(1);
+        let params = CreateSecretParams::new("expired_key", "value").with_expiry(expires_at);
+
+        store.create("user1", params).await.unwrap();
+
+        let result = store.get("user1", "expired_key").await;
+        assert!(result.is_err());
+        assert!(matches!(
+            result.unwrap_err(),
+            crate::secrets::SecretError::Expired
+        ));
+    }
+
+    #[tokio::test]
+    async fn test_non_expired_secret_succeeds() {
+        let store = test_store();
+        let expires_at = chrono::Utc::now() + chrono::Duration::hours(1);
+        let params = CreateSecretParams::new("fresh_key", "value").with_expiry(expires_at);
+
+        store.create("user1", params).await.unwrap();
+
+        let result = store.get("user1", "fresh_key").await;
+        assert!(result.is_ok());
+    }
+
     #[tokio::test]
     async fn test_user_isolation() {
         let store = test_store();

src/settings.rs

@@ -15,6 +15,10 @@ pub struct Settings {
     pub onboard_completed: bool,
 
     // === Step 1: Database ===
+    /// Database backend: "postgres" or "libsql".
+    #[serde(default)]
+    pub database_backend: Option<String>,
+
     /// Database connection URL (postgres://...).
     #[serde(default)]
     pub database_url: Option<String>,
@@ -23,13 +27,31 @@ pub struct Settings {
     #[serde(default)]
     pub database_pool_size: Option<usize>,
 
+    /// Path to local libSQL database file.
+    #[serde(default)]
+    pub libsql_path: Option<String>,
+
+    /// Turso cloud URL for remote replica sync.
+    #[serde(default)]
+    pub libsql_url: Option<String>,
+
     // === Step 2: Security ===
     /// Source for the secrets master key.
     #[serde(default)]
     pub secrets_master_key_source: KeySource,
 
-    // === Step 3: NEAR AI Auth ===
-    // Session stored separately in session.json
+    // === Step 3: Inference Provider ===
+    /// LLM backend: "nearai", "anthropic", "openai", "ollama", "openai_compatible".
+    #[serde(default)]
+    pub llm_backend: Option<String>,
+
+    /// Ollama base URL (when llm_backend = "ollama").
+    #[serde(default)]
+    pub ollama_base_url: Option<String>,
+
+    /// OpenAI-compatible endpoint base URL (when llm_backend = "openai_compatible").
+    #[serde(default)]
+    pub openai_compatible_base_url: Option<String>,
 
     // === Step 4: Model Selection ===
     /// Currently selected model.
@@ -487,20 +509,16 @@ impl Default for BuilderSettings {
 }
 
 impl Settings {
-    /// Get the default settings file path (~/.ironclaw/settings.json).
-    pub fn default_path() -> PathBuf {
-        dirs::home_dir()
-            .unwrap_or_else(|| PathBuf::from("."))
-            .join(".ironclaw")
-            .join("settings.json")
-    }
-
     /// Reconstruct Settings from a flat key-value map (as stored in the DB).
     ///
     /// Each key is a dotted path (e.g., "agent.name"), value is a JSONB value.
     /// Missing keys get their default value.
     pub fn from_db_map(map: &std::collections::HashMap<String, serde_json::Value>) -> Self {
-        // Start with defaults, then overlay each DB setting
+        // Start with defaults, then overlay each DB setting.
+        //
+        // The settings table stores both Settings struct fields and app-specific
+        // data (e.g. nearai.session_token). Skip keys that don't correspond to
+        // a known Settings path.
         let mut settings = Self::default();
 
         for (key, value) in map {
@@ -509,17 +527,23 @@ impl Settings {
                 serde_json::Value::String(s) => s.clone(),
                 serde_json::Value::Bool(b) => b.to_string(),
                 serde_json::Value::Number(n) => n.to_string(),
-                serde_json::Value::Null => "null".to_string(),
+                serde_json::Value::Null => continue, // null means default, skip
                 other => other.to_string(),
             };
 
-            if let Err(e) = settings.set(key, &value_str) {
-                tracing::warn!(
-                    "Failed to apply DB setting '{}' = '{}': {}",
-                    key,
-                    value_str,
-                    e
-                );
+            match settings.set(key, &value_str) {
+                Ok(()) => {}
+                // The settings table stores both Settings fields and app-specific
+                // data (e.g. nearai.session_token). Silently skip unknown paths.
+                Err(e) if e.starts_with("Path not found") => {}
+                Err(e) => {
+                    tracing::warn!(
+                        "Failed to apply DB setting '{}' = '{}': {}",
+                        key,
+                        value_str,
+                        e
+                    );
+                }
             }
         }
 
@@ -540,50 +564,27 @@ impl Settings {
         map
     }
 
+    /// Get the default settings file path (~/.ironclaw/settings.json).
+    pub fn default_path() -> std::path::PathBuf {
+        dirs::home_dir()
+            .unwrap_or_else(|| std::path::PathBuf::from("."))
+            .join(".ironclaw")
+            .join("settings.json")
+    }
+
     /// Load settings from disk, returning default if not found.
     pub fn load() -> Self {
         Self::load_from(&Self::default_path())
     }
 
-    /// Load settings from a specific path.
-    pub fn load_from(path: &PathBuf) -> Self {
+    /// Load settings from a specific path (used by bootstrap legacy migration).
+    pub fn load_from(path: &std::path::Path) -> Self {
         match std::fs::read_to_string(path) {
             Ok(data) => serde_json::from_str(&data).unwrap_or_default(),
             Err(_) => Self::default(),
         }
     }
 
-    /// Save settings to disk.
-    pub fn save(&self) -> std::io::Result<()> {
-        self.save_to(&Self::default_path())
-    }
-
-    /// Save settings to a specific path.
-    pub fn save_to(&self, path: &PathBuf) -> std::io::Result<()> {
-        // Ensure parent directory exists
-        if let Some(parent) = path.parent() {
-            std::fs::create_dir_all(parent)?;
-        }
-
-        let json = serde_json::to_string_pretty(self)
-            .map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?;
-
-        std::fs::write(path, json)
-    }
-
-    /// Get the selected model, falling back to the provided default.
-    pub fn model_or(&self, default: &str) -> String {
-        self.selected_model
-            .clone()
-            .unwrap_or_else(|| default.to_string())
-    }
-
-    /// Set the selected model and save.
-    pub fn set_model(&mut self, model: &str) -> std::io::Result<()> {
-        self.selected_model = Some(model.to_string());
-        self.save()
-    }
-
     /// Get a setting value by dotted path (e.g., "agent.max_parallel_jobs").
     pub fn get(&self, path: &str) -> Option<String> {
         let json = serde_json::to_value(self).ok()?;
@@ -768,42 +769,22 @@ fn collect_settings(
 #[cfg(test)]
 mod tests {
     use super::*;
-    use tempfile::tempdir;
 
     #[test]
-    fn test_settings_save_load() {
-        let dir = tempdir().unwrap();
-        let path = dir.path().join("settings.json");
-
+    fn test_db_map_round_trip() {
         let settings = Settings {
             selected_model: Some("claude-3-5-sonnet-20241022".to_string()),
             ..Default::default()
         };
 
-        settings.save_to(&path).unwrap();
-
-        let loaded = Settings::load_from(&path);
+        let map = settings.to_db_map();
+        let restored = Settings::from_db_map(&map);
         assert_eq!(
-            loaded.selected_model,
+            restored.selected_model,
             Some("claude-3-5-sonnet-20241022".to_string())
         );
     }
 
-    #[test]
-    fn test_model_or_default() {
-        let settings = Settings::default();
-        assert_eq!(
-            settings.model_or("default-model"),
-            "default-model".to_string()
-        );
-
-        let settings = Settings {
-            selected_model: Some("my-model".to_string()),
-            ..Default::default()
-        };
-        assert_eq!(settings.model_or("default-model"), "my-model".to_string());
-    }
-
     #[test]
     fn test_get_setting() {
         let settings = Settings::default();
@@ -874,16 +855,13 @@ mod tests {
     }
 
     #[test]
-    fn test_telegram_owner_id_round_trip() {
-        let dir = tempdir().unwrap();
-        let path = dir.path().join("settings.json");
-
+    fn test_telegram_owner_id_db_round_trip() {
         let mut settings = Settings::default();
         settings.channels.telegram_owner_id = Some(123456789);
-        settings.save_to(&path).unwrap();
 
-        let loaded = Settings::load_from(&path);
-        assert_eq!(loaded.channels.telegram_owner_id, Some(123456789));
+        let map = settings.to_db_map();
+        let restored = Settings::from_db_map(&map);
+        assert_eq!(restored.channels.telegram_owner_id, Some(123456789));
     }
 
     #[test]
@@ -900,4 +878,30 @@ mod tests {
             .unwrap();
         assert_eq!(settings.channels.telegram_owner_id, Some(987654321));
     }
+
+    #[test]
+    fn test_llm_backend_round_trip() {
+        let dir = tempfile::tempdir().unwrap();
+        let path = dir.path().join("settings.json");
+
+        let settings = Settings {
+            llm_backend: Some("anthropic".to_string()),
+            ollama_base_url: Some("http://localhost:11434".to_string()),
+            openai_compatible_base_url: Some("http://my-vllm:8000/v1".to_string()),
+            ..Default::default()
+        };
+        let json = serde_json::to_string_pretty(&settings).unwrap();
+        std::fs::write(&path, json).unwrap();
+
+        let loaded = Settings::load_from(&path);
+        assert_eq!(loaded.llm_backend, Some("anthropic".to_string()));
+        assert_eq!(
+            loaded.ollama_base_url,
+            Some("http://localhost:11434".to_string())
+        );
+        assert_eq!(
+            loaded.openai_compatible_base_url,
+            Some("http://my-vllm:8000/v1".to_string())
+        );
+    }
 }

src/setup/README.md

@@ -0,0 +1,539 @@
+# Setup / Onboarding Specification
+
+This document is the authoritative specification for IronClaw's onboarding
+wizard. Any code change to `src/setup/` **must** keep this document in sync.
+If a future contributor or coding agent modifies setup behavior, update this
+file first, then adjust the code to match.
+
+---
+
+## Entry Points
+
+```
+ironclaw onboard [--skip-auth] [--channels-only]
+```
+
+Explicit invocation. Loads `.env` files, runs the wizard, exits.
+
+```
+ironclaw          (first run, no database configured)
+```
+
+Auto-detection via `check_onboard_needed()` in `main.rs`. Triggers when
+none of these are true:
+- `DATABASE_URL` env var is set
+- `LIBSQL_PATH` env var is set
+- `~/.ironclaw/ironclaw.db` exists on disk
+
+The `--no-onboard` CLI flag suppresses auto-detection.
+
+---
+
+## Startup Sequence (main.rs)
+
+```
+1. Parse CLI args
+2. If Command::Onboard  → load .env, run wizard, exit
+3. If Command::Run or no command:
+   a. Load .env files (dotenvy::dotenv() then load_ironclaw_env())
+   b. check_onboard_needed() → run wizard if needed
+   c. Config::from_env()     → build config from env vars
+   d. Create SessionManager  → load session token
+   e. ensure_authenticated() → validate session (NEAR AI only)
+   f. ... rest of agent startup
+```
+
+**Critical ordering:** `.env` files must be loaded (step 3a) before
+`Config::from_env()` (step 3c) because bootstrap vars like
+`DATABASE_BACKEND` live in `~/.ironclaw/.env`.
+
+---
+
+## The 7-Step Wizard
+
+### Overview
+
+```
+Step 1: Database Connection
+Step 2: Security (master key)
+Step 3: Inference Provider          ← skipped if --skip-auth
+Step 4: Model Selection
+Step 5: Embeddings
+Step 6: Channel Configuration
+Step 7: Background Tasks (heartbeat)
+       ↓
+   save_and_summarize()
+```
+
+`--channels-only` mode runs only Step 6, skipping everything else.
+
+---
+
+### Step 1: Database Connection
+
+**Module:** `wizard.rs` → `step_database()`
+
+**Goal:** Select backend, establish connection, run migrations.
+
+**Decision tree:**
+
+```
+Both features compiled?
+├─ Yes → DATABASE_BACKEND env var set?
+│  ├─ Yes → use that backend
+│  └─ No  → interactive selection (PostgreSQL vs libSQL)
+├─ Only postgres feature → step_database_postgres()
+└─ Only libsql feature  → step_database_libsql()
+```
+
+**PostgreSQL path** (`step_database_postgres`):
+1. Check `DATABASE_URL` from env or settings
+2. Test connection (creates `deadpool_postgres::Pool`)
+3. Optionally run refinery migrations
+4. Store pool in `self.db_pool`
+
+**libSQL path** (`step_database_libsql`):
+1. Offer local path (default: `~/.ironclaw/ironclaw.db`)
+2. Optional Turso cloud sync (URL + auth token)
+3. Test connection (creates `LibSqlBackend`)
+4. Always run migrations (idempotent CREATE IF NOT EXISTS)
+5. Store backend in `self.db_backend`
+
+**Invariant:** After Step 1, exactly one of `self.db_pool` or
+`self.db_backend` is `Some`. This is required for settings persistence
+in `save_and_summarize()`.
+
+---
+
+### Step 2: Security (Master Key)
+
+**Module:** `wizard.rs` → `step_security()`
+
+**Goal:** Configure encryption for API tokens and secrets.
+
+**Decision tree:**
+
+```
+SECRETS_MASTER_KEY env var set?
+├─ Yes → use env var, done
+└─ No  → try get_master_key() from OS keychain
+   ├─ Ok(bytes) → cache in self.secrets_crypto, ask "use existing?"
+   │  ├─ Yes → done (keychain)
+   │  └─ No  → clear cache, fall through to options
+   └─ Err   → fall through to options
+              ├─ OS Keychain: generate + store + build SecretsCrypto
+              ├─ Env variable: generate + print export command
+              └─ Skip: disable secrets features
+```
+
+**CRITICAL CAVEAT: macOS Keychain Dialogs**
+
+On macOS, `security_framework::get_generic_password()` can trigger TWO
+system dialogs:
+1. "Enter your password to unlock the keychain" (keychain locked)
+2. "Allow ironclaw to access this keychain item" (per-app authorization)
+
+This is OS-level behavior we cannot prevent. To minimize pain:
+
+- **Use `get_master_key()` not `has_master_key()`** in step 2. Both call
+  the same underlying API, but `get_master_key()` returns the key bytes
+  so we can cache them. `has_master_key()` throws them away, forcing a
+  second keychain access later.
+
+- **Build `SecretsCrypto` eagerly.** When the keychain key is retrieved,
+  immediately construct `SecretsCrypto` and store in `self.secrets_crypto`.
+  Later calls to `init_secrets_context()` check this field first, avoiding
+  redundant keychain probes.
+
+- **Never probe the keychain in read-only commands** (e.g., `ironclaw status`).
+  The status command reports "env not set (keychain may be configured)"
+  rather than triggering system dialogs.
+
+**Invariant:** After Step 2, `self.secrets_crypto` is `Some` if the user
+chose Keychain or generated a new key. It may be `None` if the user chose
+env-var mode or skipped secrets.
+
+---
+
+### Step 3: Inference Provider
+
+**Module:** `wizard.rs` → `step_inference_provider()`
+
+**Goal:** Choose LLM backend and authenticate.
+
+**Providers:**
+
+| Provider | Auth Method | Secret Name | Env Var |
+|----------|-------------|-------------|---------|
+| NEAR AI | Browser OAuth | (session token) | `NEARAI_SESSION_TOKEN` |
+| Anthropic | API key | `anthropic_api_key` | `ANTHROPIC_API_KEY` |
+| OpenAI | API key | `openai_api_key` | `OPENAI_API_KEY` |
+| Ollama | None | - | - |
+| OpenAI-compatible | Optional API key | `llm_compatible_api_key` | `LLM_API_KEY` |
+
+**API-key providers** (`setup_api_key_provider`):
+1. Check env var → if set, ask to reuse, persist to secrets store
+2. Otherwise prompt for key entry via `secret_input()`
+3. Store encrypted in secrets via `init_secrets_context()`
+4. **Cache key in `self.llm_api_key`** for model fetching in Step 4
+
+**NEAR AI** (`setup_nearai`):
+- Calls `session_manager.ensure_authenticated()` which opens browser
+- Session token saved to `~/.ironclaw/session.json`
+
+**`self.llm_api_key` caching:** The wizard caches the API key as
+`Option<SecretString>` so that Step 4 (model fetching) and Step 5
+(embeddings) can use it without re-reading from the secrets store or
+mutating environment variables.
+
+---
+
+### Step 4: Model Selection
+
+**Module:** `wizard.rs` → `step_model_selection()`
+
+**Goal:** Choose which model to use.
+
+**Flow:**
+1. If model already set → offer to keep it
+2. Fetch models from provider API (5-second timeout)
+3. On timeout or error → use static fallback list
+4. Present list + "Custom model ID" escape hatch
+5. Store in `self.settings.selected_model`
+
+**Model fetchers pass the cached API key explicitly:**
+```rust
+let cached = self.llm_api_key.as_ref().map(|k| k.expose_secret().to_string());
+let models = fetch_anthropic_models(cached.as_deref()).await;
+```
+
+This avoids mutating environment variables. The fetcher checks the explicit
+key first, then falls back to the standard env var.
+
+---
+
+### Step 5: Embeddings
+
+**Module:** `wizard.rs` → `step_embeddings()`
+
+**Goal:** Configure semantic search for workspace memory.
+
+**Flow:**
+1. Ask "Enable semantic search?" (default: yes)
+2. Detect available providers:
+   - NEAR AI: if backend is `nearai` OR valid session exists
+   - OpenAI: if `OPENAI_API_KEY` in env OR (backend is `openai` AND cached key)
+3. If both available → let user choose
+4. If only one → use it
+5. If neither → disable embeddings
+
+**Default model:** `text-embedding-3-small` (for both providers)
+
+---
+
+### Step 6: Channel Configuration
+
+**Module:** `wizard.rs` → `step_channels()`, delegating to `channels.rs`
+
+**Goal:** Enable input channels (TUI, HTTP, Telegram, etc.).
+
+**Sub-steps:**
+
+```
+6a. Tunnel setup (if webhook channels needed)
+6b. Discover WASM channels from ~/.ironclaw/channels/
+6c. Multi-select: CLI/TUI, HTTP, discovered channels, bundled channels
+6d. Install missing bundled channels (copy WASM binaries)
+6e. Initialize SecretsContext (for token storage)
+6f. Setup HTTP webhook (if selected)
+6g. Setup each WASM channel (secrets, owner binding)
+```
+
+**Tunnel setup** (`setup_tunnel`):
+- Options: ngrok, Cloudflare Tunnel, localtunnel, custom URL
+- Validates HTTPS requirement
+- Stored in `self.settings.tunnel.public_url`
+
+**WASM channel setup** (`setup_wasm_channel`):
+- Reads `capabilities.json` for `setup.required_secrets`
+- For each secret: check existing, prompt or auto-generate, validate regex
+- Save each secret via `SecretsContext`
+
+**Telegram special case** (`setup_telegram`):
+- Validates bot token via Telegram `getMe` API
+- Owner binding: polls `getUpdates` for 120s to capture sender's user ID
+- Optional webhook secret generation
+
+**SecretsContext creation** (`init_secrets_context`):
+1. Check `self.secrets_crypto` (set in Step 2) → use if available
+2. Else try `SECRETS_MASTER_KEY` env var
+3. Else try `get_master_key()` from keychain (only in `channels_only` mode)
+4. Create backend-appropriate secrets store (respects selected database backend)
+
+---
+
+### Step 7: Heartbeat
+
+**Module:** `wizard.rs` → `step_heartbeat()`
+
+**Goal:** Configure periodic background execution.
+
+**Flow:**
+1. Ask "Enable heartbeat?" (default: no)
+2. If yes: interval in minutes (default: 30), notification channel
+3. Store in `self.settings.heartbeat`
+
+---
+
+## Settings Persistence
+
+### Two-Layer Architecture
+
+Settings are persisted in two places:
+
+**Layer 1: `~/.ironclaw/.env`** (bootstrap vars)
+
+Contains only the settings needed BEFORE database connection. Written by
+`save_bootstrap_env()` in `bootstrap.rs`.
+
+```env
+DATABASE_BACKEND="libsql"
+LIBSQL_PATH="/Users/name/.ironclaw/ironclaw.db"
+```
+
+Or for PostgreSQL:
+```env
+DATABASE_BACKEND="postgres"
+DATABASE_URL="postgres://user:pass@localhost/ironclaw"
+```
+
+**Why separate?** Chicken-and-egg: you need `DATABASE_BACKEND` to know
+which database to connect to, so it can't be stored in the database.
+
+**Layer 2: Database settings table** (everything else)
+
+All other settings are stored as key-value pairs in the `settings` table,
+keyed by `(user_id, key)`. Written by `set_all_settings()`.
+
+Settings are serialized via `Settings::to_db_map()` as dotted paths:
+```
+database_backend = "libsql"
+llm_backend = "nearai"
+selected_model = "anthropic/claude-sonnet-4-5"
+embeddings.enabled = "true"
+embeddings.provider = "nearai"
+channels.http_enabled = "true"
+heartbeat.enabled = "true"
+heartbeat.interval_secs = "300"
+```
+
+### save_and_summarize()
+
+Final step of the wizard:
+
+```
+1. Mark onboard_completed = true
+2. Write ALL settings to database (try postgres pool, then libSQL backend)
+3. Write bootstrap vars to ~/.ironclaw/.env:
+   - DATABASE_BACKEND (always)
+   - DATABASE_URL     (if postgres)
+   - LIBSQL_PATH      (if libsql)
+   - LIBSQL_URL       (if turso sync)
+4. Print configuration summary
+```
+
+**Invariant:** Both Layer 1 and Layer 2 must be written. If the database
+write fails, the wizard returns an error and the `.env` file is not written.
+
+### Legacy Migration
+
+`bootstrap.rs` handles one-time upgrades from older config formats:
+- `bootstrap.json` → extracts `DATABASE_URL`, writes `.env`, renames to `.migrated`
+- `settings.json` → migrated to database via `migrate_disk_to_db()`
+
+---
+
+## Settings Struct
+
+**Module:** `settings.rs`
+
+```rust
+pub struct Settings {
+    // Meta
+    pub onboard_completed: bool,
+
+    // Step 1: Database
+    pub database_backend: Option<String>,    // "postgres" | "libsql"
+    pub database_url: Option<String>,
+    pub libsql_path: Option<String>,
+    pub libsql_url: Option<String>,
+
+    // Step 2: Security
+    pub secrets_master_key_source: KeySource, // Keychain | Env | None
+
+    // Step 3: Inference
+    pub llm_backend: Option<String>,         // "nearai" | "anthropic" | "openai" | "ollama" | "openai_compatible"
+    pub ollama_base_url: Option<String>,
+    pub openai_compatible_base_url: Option<String>,
+
+    // Step 4: Model
+    pub selected_model: Option<String>,
+
+    // Step 5: Embeddings
+    pub embeddings: EmbeddingsSettings,      // enabled, provider, model
+
+    // Step 6: Channels
+    pub tunnel: TunnelSettings,              // provider, public_url
+    pub channels: ChannelSettings,           // http config, telegram owner, etc.
+
+    // Step 7: Heartbeat
+    pub heartbeat: HeartbeatSettings,        // enabled, interval, notify
+
+    // Advanced (not in wizard, set via `ironclaw config set`)
+    pub agent: AgentSettings,
+    pub wasm: WasmSettings,
+    pub sandbox: SandboxSettings,
+    pub safety: SafetySettings,
+    pub builder: BuilderSettings,
+}
+```
+
+**KeySource enum:** `Keychain | Env | None`
+
+---
+
+## Secrets Flow
+
+### SecretsContext
+
+Thin wrapper for setup-time secret operations:
+
+```rust
+pub struct SecretsContext {
+    store: Arc<dyn SecretsStore>,
+    user_id: String,
+}
+```
+
+Created by `init_secrets_context()` which:
+1. Gets `SecretsCrypto` from `self.secrets_crypto` or loads from keychain/env
+2. Creates the appropriate backend store:
+   - If both features compiled: respects `self.settings.database_backend`
+   - Tries selected backend first, falls back to the other
+3. Returns `SecretsContext` wrapping the store
+
+### Secret Storage
+
+Secrets are encrypted with AES-256-GCM using the master key, then stored
+in the database `secrets` table. The wizard writes secrets like:
+
+```
+telegram_bot_token    → encrypted bot token
+telegram_webhook_secret → encrypted webhook HMAC secret
+anthropic_api_key     → encrypted API key
+```
+
+---
+
+## Prompt Utilities
+
+**Module:** `prompts.rs`
+
+| Function | Description |
+|----------|-------------|
+| `select_one(label, options)` | Numbered single-choice menu |
+| `select_many(label, options, defaults)` | Checkbox multi-select (raw terminal mode) |
+| `input(label)` | Single line text input |
+| `optional_input(label, hint)` | Text input that can be empty |
+| `secret_input(label)` | Hidden input (shows `*` per char), returns `SecretString` |
+| `confirm(label, default)` | `[Y/n]` or `[y/N]` prompt |
+| `print_header(text)` | Bold section header with underline |
+| `print_step(n, total, text)` | `[1/7] Step Name` |
+| `print_success(text)` | Green checkmark prefix |
+| `print_error(text)` | Red X prefix |
+| `print_info(text)` | Blue info prefix |
+
+`select_many` uses `crossterm` raw mode for arrow key navigation.
+Must properly restore terminal state on all exit paths.
+
+---
+
+## Platform Caveats
+
+### macOS Keychain
+
+- `get_generic_password()` triggers system dialogs (unlock + authorize)
+- Two dialogs per call is normal, not a bug
+- Cache the result after first access to avoid repeat prompts
+- Never probe keychain in read-only commands (`status`, `--help`)
+- Service name: `"ironclaw"`, account: `"master_key"`
+
+### Linux Secret Service
+
+- Uses GNOME Keyring or KWallet via `secret-service` crate
+- May need `gnome-keyring` daemon running
+- Collection unlock may prompt for password
+
+### URL Passwords
+
+- `#` is common in URL-encoded passwords (`%23` decoded)
+- `.env` values must be double-quoted to preserve `#`
+- Display masked: `postgres://user:****@host/db`
+
+### Telegram API
+
+- Bot token format: `123456:ABC-DEF...`
+- Token goes in URL path: `https://api.telegram.org/bot{TOKEN}/method`
+- Webhook secret header: `X-Telegram-Bot-Api-Secret-Token`
+- Owner binding polls `getUpdates` (must delete webhook first)
+
+---
+
+## Testing
+
+Tests live in `mod tests {}` at the bottom of each file.
+
+**What to test when modifying setup:**
+
+- Settings round-trip: `to_db_map()` then `from_db_map()` preserves values
+- Bootstrap `.env`: dotenvy can parse what `save_bootstrap_env()` writes
+- Model fetchers: static fallback works when API is unreachable
+- Channel discovery: handles missing dir, invalid JSON, deduplication
+- Prompt functions: not tested (interactive I/O), but ensure error paths
+  don't panic
+
+**Run setup tests:**
+```bash
+cargo test --lib -- setup
+cargo test --lib -- bootstrap
+```
+
+---
+
+## Modification Checklist
+
+When changing the onboarding flow:
+
+1. Update this README first with the intended behavior change
+2. If adding a new wizard step:
+   - Add to the step enum in `run()`, adjust `total_steps`
+   - Add corresponding settings fields to `Settings`
+   - Add `to_db_map` / `from_db_map` serialization
+   - If the setting is needed before DB connection, add to `save_bootstrap_env()`
+3. If adding a new provider or channel:
+   - Add to the selection menu in the appropriate step
+   - Add authentication flow (API key or OAuth)
+   - Add model fetcher with static fallback + 5s timeout
+4. If touching keychain:
+   - Cache the result, never call `get_master_key()` twice
+   - Test on macOS (dialog behavior differs from Linux)
+5. If touching secrets:
+   - Ensure `init_secrets_context()` respects the selected database backend
+   - Test with both postgres and libsql features
+6. Run the full shipping checklist:
+   ```bash
+   cargo fmt
+   cargo clippy --all --benches --tests --examples --all-features -- -D warnings
+   cargo test --lib -- setup bootstrap
+   ```
+7. Test a fresh onboarding: `rm -rf ~/.ironclaw && cargo run`