fix: fix image preview remove button styles

chore: resolve conflicts
feat: add image+text support
2026-08-25 14:53:34 +00:00 · 2026-03-13 23:18:43 +08:00 · 2026-03-13 23:08:22 +08:00 · 2026-03-13 23:02:15 +08:00 · 2026-03-12 21:32:28 -07:00 · 2026-03-12 21:32:19 -07:00
123 changed files with 10742 additions and 1201 deletions
@@ -70,6 +70,12 @@ NEARAI_AUTH_URL=https://private.near.ai
 # LLM_BASE_URL=https://api.fireworks.ai/inference/v1
 # LLM_API_KEY=fw_...

+# === MiniMax ===
+# LLM_BACKEND=minimax
+# MINIMAX_API_KEY=...
+# MINIMAX_MODEL=MiniMax-M2.5
+# MINIMAX_BASE_URL=https://api.minimax.io/v1   # default (global); use https://api.minimaxi.com/v1 for China
+
 # === Anthropic Direct ===
 # LLM_BACKEND=anthropic
 # ANTHROPIC_MODEL=claude-sonnet-4-6
@@ -0,0 +1,23 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# Pre-push hook: run clippy and tests before pushing.
+# Install: git config core.hooksPath .githooks
+
+echo "pre-push: running clippy..."
+if ! cargo clippy --all --benches --tests --examples --all-features -- -D warnings; then
+    echo ""
+    echo "Push blocked: clippy warnings found."
+    echo "To bypass: git push --no-verify"
+    exit 1
+fi
+
+echo "pre-push: running tests..."
+if ! cargo test; then
+    echo ""
+    echo "Push blocked: tests failed."
+    echo "To bypass: git push --no-verify"
+    exit 1
+fi
+
+echo "pre-push: all checks passed."
@@ -0,0 +1,59 @@
+#!/usr/bin/env bash
+
+load_commit_summary() {
+  local range="$1"
+  local max_commits="${2:-50}"
+  local commit_list overflow
+
+  commit_list="$(git log --oneline --no-merges --reverse "${range}" 2>/dev/null || echo "")"
+  if [ -n "${commit_list}" ]; then
+    COMMIT_COUNT="$(printf '%s\n' "${commit_list}" | wc -l | tr -d ' ')"
+    if [ "${COMMIT_COUNT}" -gt "${max_commits}" ]; then
+      COMMIT_MD="$(printf '%s\n' "${commit_list}" | head -n "${max_commits}" | sed 's/^/- /')"
+      overflow=$((COMMIT_COUNT - max_commits))
+      COMMIT_MD+=$'\n'"- ... and ${overflow} more (see compare view)"
+    else
+      COMMIT_MD="$(printf '%s\n' "${commit_list}" | sed 's/^/- /')"
+    fi
+  else
+    COMMIT_COUNT=0
+    COMMIT_MD="- (no non-merge commits in range)"
+  fi
+}
+
+replace_marked_section() {
+  local body_file="$1"
+  local section_file="$2"
+  local section_start="$3"
+  local section_end="$4"
+  local output_file="$5"
+
+  if grep -qF "${section_start}" "${body_file}" && grep -qF "${section_end}" "${body_file}"; then
+    awk -v start="${section_start}" -v end="${section_end}" -v replacement_file="${section_file}" '
+      BEGIN {
+        while ((getline line < replacement_file) > 0) {
+          replacement = replacement line ORS
+        }
+        in_block = 0
+      }
+      $0 == start {
+        printf "%s", replacement
+        in_block = 1
+        next
+      }
+      $0 == end {
+        in_block = 0
+        next
+      }
+      !in_block {
+        print
+      }
+    ' "${body_file}" > "${output_file}"
+  else
+    cp "${body_file}" "${output_file}"
+    if [ -s "${output_file}" ]; then
+      printf '\n\n' >> "${output_file}"
+    fi
+    cat "${section_file}" >> "${output_file}"
+  fi
+}
@@ -0,0 +1,101 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+: "${PR_NUMBER:?PR_NUMBER is required}"
+: "${REPO:?REPO is required}"
+
+MAIN_BRANCH="${MAIN_BRANCH:-main}"
+DRY_RUN="${DRY_RUN:-false}"
+SECTION_START="<!-- staging-promotion-release-summary:start -->"
+SECTION_END="<!-- staging-promotion-release-summary:end -->"
+TMP_DIR="$(mktemp -d)"
+trap 'rm -rf "${TMP_DIR}"' EXIT
+
+# shellcheck source=.github/scripts/pr-body-utils.sh
+source "$(dirname "$0")/pr-body-utils.sh"
+
+gh pr view "${PR_NUMBER}" --repo "${REPO}" --json body > "${TMP_DIR}/pr.json"
+jq -r '.body // ""' < "${TMP_DIR}/pr.json" > "${TMP_DIR}/body.md"
+
+git fetch origin "${MAIN_BRANCH}"
+git fetch origin "+refs/tags/v*:refs/tags/v*"
+
+LAST_TAG="$(git describe --tags --match 'v*' --abbrev=0 "origin/${MAIN_BRANCH}" 2>/dev/null || true)"
+if [ -n "${LAST_TAG}" ]; then
+  RANGE="${LAST_TAG}..origin/${MAIN_BRANCH}"
+  HEADER="## Staging promotion batches since ${LAST_TAG}"
+  EMPTY_MESSAGE="_No structured staging promotion merges found since ${LAST_TAG}._"
+else
+  RANGE="origin/${MAIN_BRANCH}"
+  HEADER="## Staging promotion batches on ${MAIN_BRANCH}"
+  EMPTY_MESSAGE="_No structured staging promotion merges found on ${MAIN_BRANCH}._"
+fi
+
+{
+  echo "${SECTION_START}"
+  echo "${HEADER}"
+  echo
+} > "${TMP_DIR}/section.md"
+
+FOUND_SUMMARY=false
+while IFS= read -r sha; do
+  [ -n "${sha}" ] || continue
+  BODY="$(git show -s --format=%b "${sha}")"
+  if ! printf '%s\n' "${BODY}" | grep -q '^staging-promotion-summary-v1$'; then
+    continue
+  fi
+
+  FOUND_SUMMARY=true
+  SUBJECT="$(git show -s --format=%s "${sha}")"
+  PR_REF="$(printf '%s\n' "${BODY}" | sed -n 's/^promotion-pr: //p' | head -n 1)"
+  COMMIT_COUNT="$(printf '%s\n' "${BODY}" | sed -n 's/^current-commit-count: //p' | head -n 1)"
+  CURRENT_RANGE="$(printf '%s\n' "${BODY}" | sed -n 's/^current-range: //p' | head -n 1)"
+  COMMIT_BLOCK="$(printf '%s\n' "${BODY}" | awk 'capture { print } /^Current commits in this promotion \([0-9]+\):$/ { capture = 1 }')"
+
+  {
+    echo "### ${SUBJECT}"
+    echo
+    if [ -n "${PR_REF}" ]; then
+      echo "**Promotion PR:** ${PR_REF}"
+    fi
+    if [ -n "${COMMIT_COUNT}" ]; then
+      echo "**Commit count:** ${COMMIT_COUNT}"
+    fi
+    if [ -n "${CURRENT_RANGE}" ]; then
+      echo "**Range:** \`${CURRENT_RANGE}\`"
+    fi
+    echo
+    if [ -n "${COMMIT_BLOCK}" ]; then
+      echo "${COMMIT_BLOCK}"
+    else
+      echo "- (no commit summary found)"
+    fi
+    echo
+  } >> "${TMP_DIR}/section.md"
+done < <(git log --merges --reverse --format='%H' "${RANGE}")
+
+if [ "${FOUND_SUMMARY}" = false ]; then
+  {
+    echo "${EMPTY_MESSAGE}"
+    echo
+  } >> "${TMP_DIR}/section.md"
+fi
+
+{
+  echo "*Auto-updated from structured staging promotion merge bodies on ${MAIN_BRANCH}.*"
+  echo "${SECTION_END}"
+} >> "${TMP_DIR}/section.md"
+
+replace_marked_section \
+  "${TMP_DIR}/body.md" \
+  "${TMP_DIR}/section.md" \
+  "${SECTION_START}" \
+  "${SECTION_END}" \
+  "${TMP_DIR}/new-body.md"
+
+if [ "${DRY_RUN}" = "true" ]; then
+  echo "Dry run enabled. Computed PR body for #${PR_NUMBER}:"
+  cat "${TMP_DIR}/new-body.md"
+else
+  gh pr edit "${PR_NUMBER}" --repo "${REPO}" --body-file "${TMP_DIR}/new-body.md"
+fi
@@ -0,0 +1,53 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+: "${PR_NUMBER:?PR_NUMBER is required}"
+: "${REPO:?REPO is required}"
+
+MAX_COMMITS="${MAX_COMMITS:-50}"
+DRY_RUN="${DRY_RUN:-false}"
+SECTION_START="<!-- staging-ci-current:start -->"
+SECTION_END="<!-- staging-ci-current:end -->"
+TMP_DIR="$(mktemp -d)"
+trap 'rm -rf "${TMP_DIR}"' EXIT
+
+# shellcheck source=.github/scripts/pr-body-utils.sh
+source "$(dirname "$0")/pr-body-utils.sh"
+
+gh pr view "${PR_NUMBER}" --repo "${REPO}" --json body,baseRefName,headRefName > "${TMP_DIR}/pr.json"
+jq -r '.body // ""' < "${TMP_DIR}/pr.json" > "${TMP_DIR}/body.md"
+BASE="$(jq -r '.baseRefName' < "${TMP_DIR}/pr.json")"
+HEAD="$(jq -r '.headRefName' < "${TMP_DIR}/pr.json")"
+RANGE="origin/${BASE}..origin/${HEAD}"
+
+git fetch origin "${BASE}" "${HEAD}"
+
+load_commit_summary "${RANGE}" "${MAX_COMMITS}"
+
+{
+  echo "${SECTION_START}"
+  echo "### Current commits in this promotion (${COMMIT_COUNT})"
+  echo
+  echo "**Current base:** \`${BASE}\`"
+  echo "**Current head:** \`${HEAD}\`"
+  echo "**Current range:** \`${RANGE}\`"
+  echo
+  echo "${COMMIT_MD}"
+  echo
+  echo "*Auto-updated by staging promotion metadata workflow*"
+  echo "${SECTION_END}"
+} > "${TMP_DIR}/section.md"
+
+replace_marked_section \
+  "${TMP_DIR}/body.md" \
+  "${TMP_DIR}/section.md" \
+  "${SECTION_START}" \
+  "${SECTION_END}" \
+  "${TMP_DIR}/new-body.md"
+
+if [ "${DRY_RUN}" = "true" ]; then
+  echo "Dry run enabled. Computed PR body for #${PR_NUMBER}:"
+  cat "${TMP_DIR}/new-body.md"
+else
+  gh pr edit "${PR_NUMBER}" --repo "${REPO}" --body-file "${TMP_DIR}/new-body.md"
+fi
@@ -48,11 +48,11 @@ jobs:
      matrix:
        include:
          - group: core
-            files: "tests/e2e/scenarios/test_connection.py tests/e2e/scenarios/test_chat.py tests/e2e/scenarios/test_sse_reconnect.py tests/e2e/scenarios/test_html_injection.py"
+            files: "tests/e2e/scenarios/test_connection.py tests/e2e/scenarios/test_chat.py tests/e2e/scenarios/test_sse_reconnect.py tests/e2e/scenarios/test_html_injection.py tests/e2e/scenarios/test_csp.py"
          - group: features
            files: "tests/e2e/scenarios/test_skills.py tests/e2e/scenarios/test_tool_approval.py"
          - group: extensions
-            files: "tests/e2e/scenarios/test_extensions.py"
+            files: "tests/e2e/scenarios/test_extensions.py tests/e2e/scenarios/test_extension_oauth.py tests/e2e/scenarios/test_wasm_lifecycle.py tests/e2e/scenarios/test_tool_execution.py tests/e2e/scenarios/test_pairing.py"
    steps:
      - uses: actions/checkout@v6

@@ -13,6 +13,11 @@ jobs:
        with:
          fetch-depth: 0

+      - name: Fetch PR head and base
+        run: |
+          git fetch origin ${{ github.event.pull_request.base.ref }}
+          git fetch origin pull/${{ github.event.pull_request.number }}/head:pr-head
+
      - name: Check for regression tests
        env:
          PR_TITLE: ${{ github.event.pull_request.title }}
@@ -21,6 +26,8 @@ jobs:
          set -euo pipefail

          BASE_REF="origin/${{ github.event.pull_request.base.ref }}"
+          # Use the actual PR head, not the merge commit that actions/checkout checks out
+          HEAD_REF="pr-head"

          # --- 1. Is this a fix PR? Check title first, then commit messages ---
          IS_FIX=false
@@ -30,7 +37,7 @@ jobs:
          fi

          if [ "$IS_FIX" = false ]; then
-            COMMITS=$(git log --format='%s' "${BASE_REF}..HEAD")
+            COMMITS=$(git log --format='%s' "${BASE_REF}..${HEAD_REF}")
            if grep -qiE '^(fix(\(.*\))?|hotfix|bugfix):' <<< "$COMMITS"; then
              IS_FIX=true
            fi
@@ -49,14 +56,14 @@ jobs:
            exit 0
          fi

-          COMMIT_BODIES=$(git log --format='%B' "${BASE_REF}..HEAD")
+          COMMIT_BODIES=$(git log --format='%B' "${BASE_REF}..${HEAD_REF}")
          if grep -qF '[skip-regression-check]' <<< "$COMMIT_BODIES"; then
            echo "[skip-regression-check] found in commit message — skipping."
            exit 0
          fi

          # --- 3. Exempt static-only / docs-only changes ---
-          CHANGED_FILES=$(git diff --name-only "${BASE_REF}...HEAD")
+          CHANGED_FILES=$(git diff --name-only "${BASE_REF}...${HEAD_REF}")

          if [ -z "$CHANGED_FILES" ]; then
            echo "No changed files — skipping."
@@ -80,13 +87,13 @@ jobs:
          # --- 4. Look for test changes ---

          # Fast path: new test attributes or test modules in added lines.
-          if git diff "${BASE_REF}...HEAD" -U0 -- '*.rs' | grep -qE '^\+.*(#\[test\]|#\[tokio::test\]|#\[cfg\(test\)\]|mod tests)'; then
+          if git diff "${BASE_REF}...${HEAD_REF}" -U0 -- '*.rs' | grep -qE '^\+.*(#\[test\]|#\[tokio::test\]|#\[cfg\(test\)\]|mod tests)'; then
            echo "Test changes found in .rs files."
            exit 0
          fi

          # Whole-function context: detect edits inside existing test functions.
-          if git diff "${BASE_REF}...HEAD" -W -- '*.rs' | awk '
+          if git diff "${BASE_REF}...${HEAD_REF}" -W -- '*.rs' | awk '
            /^@@/           { if (has_test && has_add) { found=1; exit } has_test=0; has_add=0 }
            /^ .*#\[test\]/ || /^ .*#\[tokio::test\]/ || /^ .*#\[cfg\(test\)\]/ || /^ .*mod tests/ { has_test=1 }
            /^\+.*#\[test\]/ || /^\+.*#\[tokio::test\]/ || /^\+.*#\[cfg\(test\)\]/ || /^\+.*mod tests/ { has_test=1 }
@@ -0,0 +1,44 @@
+name: Release-plz Batch Summary
+
+on:
+  workflow_dispatch:
+    inputs:
+      pr_number:
+        description: "release-plz PR number to refresh"
+        required: true
+        type: string
+      dry_run:
+        description: "Compute the body update without editing the PR"
+        required: false
+        type: boolean
+        default: true
+  pull_request_target:
+    types: [opened, synchronize, reopened]
+
+permissions:
+  contents: read
+  pull-requests: write
+
+jobs:
+  update-release-pr:
+    if: >
+      (github.event_name == 'pull_request_target' &&
+       github.event.pull_request.head.repo.full_name == github.repository &&
+       startsWith(github.event.pull_request.head.ref, 'release-plz-')) ||
+      github.event_name == 'workflow_dispatch'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout base branch
+        uses: actions/checkout@v6
+        with:
+          ref: ${{ github.event_name == 'workflow_dispatch' && 'main' || github.event.pull_request.base.ref }}
+          fetch-depth: 0
+          fetch-tags: true
+
+      - name: Update release-plz PR body with staging batch summary
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          PR_NUMBER: ${{ github.event_name == 'workflow_dispatch' && inputs.pr_number || github.event.pull_request.number }}
+          REPO: ${{ github.repository }}
+          DRY_RUN: ${{ github.event_name == 'workflow_dispatch' && inputs.dry_run || 'false' }}
+        run: bash .github/scripts/update-release-plz-body.sh
@@ -58,10 +58,16 @@ jobs:
      - *checkout
      - *install-rust
      - uses: Swatinem/rust-cache@v2
+      - name: Generate GitHub token
+        uses: actions/create-github-app-token@v2
+        id: generate-token
+        with:
+          app-id: ${{ secrets.GH_RELEASES_MANAGER_APP_ID }}
+          private-key: ${{ secrets.GH_RELEASES_MANAGER_APP_PRIVATE_KEY }}
      - name: Run release-plz
        uses: release-plz/[email protected]
        with:
          command: release-pr
        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          GITHUB_TOKEN: ${{ steps.generate-token.outputs.token }}
          CARGO_REGISTRY_TOKEN: ${{ secrets.CARGO_REGISTRY_TOKEN }}
@@ -25,9 +25,35 @@ concurrency:
  cancel-in-progress: false        # Let running suites finish

 jobs:
+  # ── Resolve promotion base branch ───────────────────────────────
+  resolve-promotion-base:
+    name: Resolve promotion base
+    runs-on: ubuntu-latest
+    outputs:
+      promotion_base: ${{ steps.resolve.outputs.promotion_base }}
+    steps:
+      - name: Resolve promotion base
+        id: resolve
+        env:
+          GH_TOKEN: ${{ github.token }}
+          FALLBACK_BRANCH: main
+          REPO: ${{ github.repository }}
+        run: |
+          LATEST=$(gh pr list --repo "${REPO}" --label staging-promotion --state open \
+            --json headRefName,createdAt \
+            --jq '[.[] | select(.headRefName | startswith("staging-promote/"))] | sort_by(.createdAt) | last | .headRefName // empty')
+          if [ -n "$LATEST" ]; then
+            echo "promotion_base=${LATEST}" >> "$GITHUB_OUTPUT"
+            echo "Using open promotion branch as base: ${LATEST}"
+          else
+            echo "promotion_base=${FALLBACK_BRANCH}" >> "$GITHUB_OUTPUT"
+            echo "No open promotion branch found. Using ${FALLBACK_BRANCH}."
+          fi
+
  # ── Check for new commits ──────────────────────────────────────
  check-changes:
    name: Check for new commits
+    needs: resolve-promotion-base
    runs-on: ubuntu-latest
    outputs:
      has_changes: ${{ steps.check.outputs.has_changes }}
@@ -44,6 +70,7 @@ jobs:
        id: check
        env:
          FORCE_RUN: ${{ inputs.force }}
+          PROMOTION_BASE: ${{ needs.resolve-promotion-base.outputs.promotion_base }}
        run: |
          CURRENT_HEAD=$(git rev-parse HEAD)
          echo "current_head=${CURRENT_HEAD}" >> "$GITHUB_OUTPUT"
@@ -65,9 +92,9 @@ jobs:
              echo "Found ${COMMIT_COUNT} new commit(s) since last tested"
              DIFF_RANGE="${LAST_TESTED}..${CURRENT_HEAD}"
            else
-              git fetch origin main
-              MERGE_BASE=$(git merge-base origin/main HEAD)
-              echo "First run -- reviewing from merge-base ${MERGE_BASE}"
+              git fetch origin "${PROMOTION_BASE}"
+              MERGE_BASE=$(git merge-base "origin/${PROMOTION_BASE}" HEAD)
+              echo "First run -- reviewing from merge-base ${MERGE_BASE} against ${PROMOTION_BASE}"
              DIFF_RANGE="${MERGE_BASE}..${CURRENT_HEAD}"
            fi
          fi
@@ -101,7 +128,7 @@ jobs:
  # ── Create promotion PR (triggers claude-review.yml on the PR) ──
  create-promotion-pr:
    name: Create Promotion PR
-    needs: check-changes
+    needs: [resolve-promotion-base, check-changes]
    if: needs.check-changes.outputs.has_changes == 'true'
    runs-on: ubuntu-latest
    outputs:
@@ -129,18 +156,19 @@ jobs:
            echo "token=${{ github.token }}" >> "$GITHUB_OUTPUT"
          fi

-      - name: Check if staging is ahead of main
+      - name: Check if staging is ahead of target branch
        id: ahead-check
        env:
          GH_TOKEN: ${{ steps.token.outputs.token }}
+          PROMOTION_BASE: ${{ needs.resolve-promotion-base.outputs.promotion_base }}
        run: |
-          git fetch origin main
-          AHEAD=$(git rev-list --count origin/main..origin/staging)
+          git fetch origin "${PROMOTION_BASE}"
+          AHEAD=$(git rev-list --count "origin/${PROMOTION_BASE}..origin/staging")
          echo "commits_ahead=${AHEAD}" >> "$GITHUB_OUTPUT"
          if [ "$AHEAD" -eq 0 ]; then
-            echo "Staging is not ahead of main. Nothing to promote."
+            echo "Staging is not ahead of ${PROMOTION_BASE}. Nothing to promote."
          else
-            echo "Staging is ${AHEAD} commits ahead of main."
+            echo "Staging is ${AHEAD} commits ahead of ${PROMOTION_BASE}."
          fi

      - name: Create promotion branch
@@ -154,53 +182,53 @@ jobs:
          echo "branch=${BRANCH}" >> "$GITHUB_OUTPUT"
          echo "Created promotion branch: ${BRANCH}"

-      - name: Find base branch
-        id: find-base
-        if: steps.ahead-check.outputs.commits_ahead != '0'
-        env:
-          GH_TOKEN: ${{ steps.token.outputs.token }}
-        run: |
-          # Find the newest open promotion PR with a staging-promote/* head branch
-          LATEST=$(gh pr list --label staging-promotion --state open \
-            --json headRefName,createdAt \
-            --jq '[.[] | select(.headRefName | startswith("staging-promote/"))] | sort_by(.createdAt) | last | .headRefName // empty')
-          if [ -n "$LATEST" ]; then
-            echo "base=${LATEST}" >> "$GITHUB_OUTPUT"
-            echo "Chaining onto existing promotion branch: ${LATEST}"
-          else
-            echo "base=main" >> "$GITHUB_OUTPUT"
-            echo "No existing promotion PR — targeting main"
-          fi
-
      - name: Create promotion PR
        id: create-pr
        if: steps.ahead-check.outputs.commits_ahead != '0'
        env:
          GH_TOKEN: ${{ steps.token.outputs.token }}
        run: |
+          source .github/scripts/pr-body-utils.sh
          RANGE="${{ needs.check-changes.outputs.diff_range }}"
          TIMESTAMP=$(date -u +"%Y-%m-%d %H:%M UTC")
          BRANCH="${{ steps.branch.outputs.branch }}"
-          BASE="${{ steps.find-base.outputs.base }}"
+          BASE="${{ needs.resolve-promotion-base.outputs.promotion_base }}"
+
+          MAX_COMMITS=50
+          load_commit_summary "${RANGE}" "${MAX_COMMITS}"
+
+          # Build PR body via concatenation to avoid heredoc shell expansion
+          # (commit messages in COMMIT_MD may contain $, backticks, or backslashes)
+          PR_BODY="## Auto-promotion from staging CI"
+          PR_BODY+=$'\n\n'"**Batch range:** \`${RANGE}\`"
+          PR_BODY+=$'\n'"**Promotion branch:** \`${BRANCH}\`"
+          PR_BODY+=$'\n'"**Base:** \`${BASE}\`"
+          PR_BODY+=$'\n'"**Triggered by:** Staging CI batch at ${TIMESTAMP}"
+          PR_BODY+=$'\n\n'"### Commits in this batch (${COMMIT_COUNT}):"
+          PR_BODY+=$'\n'"${COMMIT_MD}"
+          PR_BODY+=$'\n\n'"<!-- staging-ci-current:start -->"
+          PR_BODY+=$'\n'"### Current commits in this promotion (${COMMIT_COUNT})"
+          PR_BODY+=$'\n'
+          PR_BODY+=$'\n'"**Current base:** \`${BASE}\`"
+          PR_BODY+=$'\n'"**Current head:** \`${BRANCH}\`"
+          PR_BODY+=$'\n'"**Current range:** \`origin/${BASE}..origin/${BRANCH}\`"
+          PR_BODY+=$'\n'
+          PR_BODY+=$'\n'"${COMMIT_MD}"
+          PR_BODY+=$'\n'
+          PR_BODY+=$'\n'"*Auto-updated by staging promotion metadata workflow*"
+          PR_BODY+=$'\n'"<!-- staging-ci-current:end -->"
+          PR_BODY+=$'\n\n'"Waiting for gates:"
+          PR_BODY+=$'\n'"- Tests: pending"
+          PR_BODY+=$'\n'"- E2E: pending"
+          PR_BODY+=$'\n'"- Claude Code review: pending (will post comments on this PR)"
+          PR_BODY+=$'\n\n'"---"
+          PR_BODY+=$'\n'"*Auto-created by staging-ci workflow*"

          PR_URL=$(gh pr create \
            --base "$BASE" \
            --head "$BRANCH" \
-            --title "chore: promote staging to main (${TIMESTAMP})" \
-            --body "## Auto-promotion from staging CI
-
-          **Batch range:** \`${RANGE}\`
-          **Promotion branch:** \`${BRANCH}\`
-          **Base:** \`${BASE}\`
-          **Triggered by:** Staging CI batch at ${TIMESTAMP}
-
-          Waiting for gates:
-          - Tests: pending
-          - E2E: pending
-          - Claude Code review: pending (will post comments on this PR)
-
-          ---
-          *Auto-created by staging-ci workflow*" \
+            --title "chore: promote staging to ${BASE} (${TIMESTAMP})" \
+            --body "$PR_BODY" \
            --label "staging-promotion")

          PR_NUM=$(echo "$PR_URL" | grep -oE '[0-9]+$')
@@ -225,7 +253,8 @@ jobs:
      - uses: actions/checkout@v6
        with:
          ref: staging
-          fetch-depth: 1
+          # Need full history to recompute the final promoted range before merge.
+          fetch-depth: 0

      - name: Generate GitHub App token
        id: app-token
@@ -324,8 +353,10 @@ jobs:
          # Use process substitution so variables propagate to parent shell
          while read -r line; do
            TAG=$(echo "$line" | grep -oE '^\[(CRITICAL|HIGH|MEDIUM|LOW):[0-9]+\]')
-            SEVERITY=$(echo "$TAG" | sed 's/\[\(.*\):\(.*\)\]/\1/')
-            CONFIDENCE=$(echo "$TAG" | sed 's/\[\(.*\):\(.*\)\]/\2/')
+            SEVERITY="${TAG#\[}"
+            SEVERITY="${SEVERITY%%:*}"
+            CONFIDENCE="${TAG##*:}"
+            CONFIDENCE="${CONFIDENCE%\]}"
            DESC=$(echo "$line" | sed "s/\[${SEVERITY}:${CONFIDENCE}\] *//" | head -1)

            echo "Found: [${SEVERITY}:${CONFIDENCE}] ${DESC}"
@@ -417,11 +448,29 @@ jobs:
          GH_TOKEN: ${{ steps.token.outputs.token }}
          PR_NUMBER: ${{ needs.create-promotion-pr.outputs.pr_number }}
        run: |
+          source .github/scripts/pr-body-utils.sh
          if [ -n "$PR_NUMBER" ]; then
            BASE=$(gh pr view "$PR_NUMBER" --json baseRefName --jq '.baseRefName')
            if [ "$BASE" = "main" ]; then
              echo "Merging promotion PR #${PR_NUMBER} (targets main)"
-              gh pr merge "$PR_NUMBER" --merge
+              TITLE=$(gh pr view "$PR_NUMBER" --json title --jq '.title')
+              HEAD_BRANCH=$(gh pr view "$PR_NUMBER" --json headRefName --jq '.headRefName')
+              git fetch origin "${BASE}" "${HEAD_BRANCH}"
+              CURRENT_RANGE="origin/${BASE}..origin/${HEAD_BRANCH}"
+              MAX_COMMITS=50
+              load_commit_summary "${CURRENT_RANGE}" "${MAX_COMMITS}"
+              {
+                echo "staging-promotion-summary-v1"
+                echo "promotion-pr: #${PR_NUMBER}"
+                echo "base: ${BASE}"
+                echo "head: ${HEAD_BRANCH}"
+                echo "current-range: ${CURRENT_RANGE}"
+                echo "current-commit-count: ${COMMIT_COUNT}"
+                echo ""
+                echo "Current commits in this promotion (${COMMIT_COUNT}):"
+                echo "${COMMIT_MD}"
+              } > /tmp/staging-promotion-merge-body.md
+              gh pr merge "$PR_NUMBER" --merge --subject "#${PR_NUMBER} $TITLE" --body-file /tmp/staging-promotion-merge-body.md
              echo "merged=true" >> "$GITHUB_OUTPUT"
            else
              echo "PR #${PR_NUMBER} targets '${BASE}' (not main) — leaving open for chain resolution"
@@ -461,18 +510,20 @@ jobs:
    steps:
      - name: Summary
        run: |
-          echo "## Staging CI Batch Results" >> "$GITHUB_STEP_SUMMARY"
-          echo "" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Check | Result |" >> "$GITHUB_STEP_SUMMARY"
-          echo "|-------|--------|" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Tests | ${{ needs.tests.result }} |" >> "$GITHUB_STEP_SUMMARY"
-          echo "| E2E | ${{ needs.e2e.result }} |" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Promotion PR | ${{ needs.create-promotion-pr.result }} |" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Gate | ${{ needs.gate.result }} |" >> "$GITHUB_STEP_SUMMARY"
-          echo "| Tag Updated | ${{ needs.update-tag.result }} |" >> "$GITHUB_STEP_SUMMARY"
-          echo "" >> "$GITHUB_STEP_SUMMARY"
-          echo "Range: ${{ needs.check-changes.outputs.diff_range }}" >> "$GITHUB_STEP_SUMMARY"
-          PR_NUM="${{ needs.create-promotion-pr.outputs.pr_number }}"
-          if [ -n "$PR_NUM" ]; then
-            echo "Promotion PR: #${PR_NUM}" >> "$GITHUB_STEP_SUMMARY"
-          fi
+          {
+            echo "## Staging CI Batch Results"
+            echo ""
+            echo "| Check | Result |"
+            echo "|-------|--------|"
+            echo "| Tests | ${{ needs.tests.result }} |"
+            echo "| E2E | ${{ needs.e2e.result }} |"
+            echo "| Promotion PR | ${{ needs.create-promotion-pr.result }} |"
+            echo "| Gate | ${{ needs.gate.result }} |"
+            echo "| Tag Updated | ${{ needs.update-tag.result }} |"
+            echo ""
+            echo "Range: ${{ needs.check-changes.outputs.diff_range }}"
+            PR_NUM="${{ needs.create-promotion-pr.outputs.pr_number }}"
+            if [ -n "$PR_NUM" ]; then
+              echo "Promotion PR: #${PR_NUM}"
+            fi
+          } >> "$GITHUB_STEP_SUMMARY"
@@ -0,0 +1,78 @@
+name: Staging Promotion Metadata
+
+on:
+  workflow_dispatch:
+    inputs:
+      pr_number:
+        description: "Staging promotion PR number to refresh"
+        required: true
+        type: string
+      dry_run:
+        description: "Compute the body update without editing the PR"
+        required: false
+        type: boolean
+        default: true
+  pull_request_target:
+    types: [opened, synchronize, reopened]
+  push:
+    branches:
+      - main
+
+permissions:
+  contents: read
+  pull-requests: write
+
+jobs:
+  refresh-single-pr:
+    if: >
+      (github.event_name == 'pull_request_target' &&
+       github.event.pull_request.head.repo.full_name == github.repository &&
+       startsWith(github.event.pull_request.head.ref, 'staging-promote/')) ||
+      github.event_name == 'workflow_dispatch'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout workflow source
+        uses: actions/checkout@v6
+        with:
+          # For chained promotion PRs, the script lives on the trusted PR head,
+          # not necessarily on the older promotion branch used as the PR base.
+          ref: ${{ github.event_name == 'workflow_dispatch' && 'main' || github.event.pull_request.head.sha }}
+          fetch-depth: 0
+          fetch-tags: true
+
+      - name: Refresh staging promotion PR body
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          PR_NUMBER: ${{ github.event_name == 'workflow_dispatch' && inputs.pr_number || github.event.pull_request.number }}
+          REPO: ${{ github.repository }}
+          DRY_RUN: ${{ github.event_name == 'workflow_dispatch' && inputs.dry_run || 'false' }}
+        run: bash .github/scripts/update-staging-promotion-body.sh
+
+  refresh-open-prs-after-main-push:
+    if: github.event_name == 'push'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout main
+        uses: actions/checkout@v6
+        with:
+          ref: main
+          fetch-depth: 0
+          fetch-tags: true
+
+      - name: Refresh all open staging promotion PR bodies
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          REPO: ${{ github.repository }}
+        run: |
+          # ubuntu-latest uses bash 5.x, so mapfile is available here.
+          mapfile -t prs < <(gh pr list --repo "${REPO}" --label staging-promotion --state open \
+            --json number,headRefName \
+            --jq '.[] | select(.headRefName | startswith("staging-promote/")) | .number')
+          if [ "${#prs[@]}" -eq 0 ]; then
+            echo "No open staging promotion PRs to refresh."
+            exit 0
+          fi
+          for pr in "${prs[@]}"; do
+            echo "Refreshing staging promotion PR #${pr}"
+            PR_NUMBER="${pr}" bash .github/scripts/update-staging-promotion-body.sh
+          done
@@ -7,6 +7,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

 ## [Unreleased]

+## [0.18.0](https://github.com/nearai/ironclaw/compare/v0.17.0...v0.18.0) - 2026-03-11
+
+### Other
+
+- Merge pull request #907 from nearai/staging-promote/b0214fef-22930316561
+- promote staging to main (2026-03-10 15:19 UTC) ([#865](https://github.com/nearai/ironclaw/pull/865))
+- Merge pull request #830 from nearai/staging-promote/3a2989d0-22888378864
+- update WASM artifact SHA256 checksums [skip ci] ([#876](https://github.com/nearai/ironclaw/pull/876))
+
 ## [0.17.0](https://github.com/nearai/ironclaw/compare/v0.16.1...v0.17.0) - 2026-03-10

 ### Added
@@ -3350,7 +3350,7 @@ dependencies = [

 [[package]]
 name = "ironclaw"
-version = "0.17.0"
+version = "0.18.0"
 dependencies = [
 "aes-gcm",
 "aho-corasick",
@@ -20,7 +20,7 @@ exclude = [

 [package]
 name = "ironclaw"
-version = "0.17.0"
+version = "0.18.0"
 edition = "2024"
 rust-version = "1.92"
 description = "Secure personal AI assistant that protects your data and expands its capabilities on the fly"
@@ -19,6 +19,7 @@ WORKDIR /app

 # Copy manifests first for layer caching
 COPY Cargo.toml Cargo.lock ./
+COPY crates/ crates/

 # Copy source, build script, tests, and supporting directories
 COPY build.rs build.rs
@@ -20,6 +20,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 WORKDIR /app

 COPY Cargo.toml Cargo.lock ./
+COPY crates/ crates/
 COPY build.rs build.rs
 COPY src/ src/
 COPY tests/ tests/
@@ -170,7 +170,7 @@ This document tracks feature parity between IronClaw (Rust implementation) and O
 | `nodes` | ✅ | ❌ | P3 | Device management, remove/clear flows |
 | `plugins` | ✅ | ❌ | P3 | Plugin management |
 | `hooks` | ✅ | ✅ | P2 | Lifecycle hooks |
-| `cron` | ✅ | ❌ | P2 | Scheduled jobs (model/thinking fields in edit) |
+| `cron` | ✅ | 🚧 | P2 | list/create/edit/enable/disable/delete/history; TODO: `cron run`, model/thinking fields |
 | `webhooks` | ✅ | ❌ | P3 | Webhook config |
 | `message send` | ✅ | ❌ | P2 | Send to channels |
 | `browser` | ✅ | ❌ | P3 | Browser automation |
@@ -20,12 +20,27 @@ version = "1.0.102"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7f202df86484c868dbad7eaa557ef785d5c66295e41b460ef922eca0723b842c"

+[[package]]
+name = "base64ct"
+version = "1.8.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2af50177e190e07a26ab74f8b1efbfe2ef87da2116221318cb1c2e82baf7de06"
+
 [[package]]
 name = "bitflags"
 version = "2.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "843867be96c8daad0d758b57df9392b6d8d271134fce549de6ce169ff98a92af"

+[[package]]
+name = "block-buffer"
+version = "0.10.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3078c7629b62d3f0439517fa394996acacc5cbc91c5a20d8c658e77abd503a71"
+dependencies = [
+ "generic-array",
+]
+
 [[package]]
 name = "cfg-if"
 version = "1.0.4"
@@ -33,20 +48,134 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"

 [[package]]
-name = "discord-channel"
-version = "0.1.0"
+name = "const-oid"
+version = "0.9.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c2459377285ad874054d797f3ccebf984978aa39129f6eafde5cdc8315b612f8"
+
+[[package]]
+name = "cpufeatures"
+version = "0.2.17"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "59ed5838eebb26a2bb2e58f6d5b5316989ae9d08bab10e0e6d103e656d1b0280"
 dependencies = [
+ "libc",
+]
+
+[[package]]
+name = "crypto-common"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "78c8292055d1c1df0cce5d180393dc8cce0abec0a7102adb6c7b1eef6016d60a"
+dependencies = [
+ "generic-array",
+ "typenum",
+]
+
+[[package]]
+name = "curve25519-dalek"
+version = "4.1.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "97fb8b7c4503de7d6ae7b42ab72a5a59857b4c937ec27a3d4539dba95b5ab2be"
+dependencies = [
+ "cfg-if",
+ "cpufeatures",
+ "curve25519-dalek-derive",
+ "digest",
+ "fiat-crypto",
+ "rustc_version",
+ "subtle",
+ "zeroize",
+]
+
+[[package]]
+name = "curve25519-dalek-derive"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f46882e17999c6cc590af592290432be3bce0428cb0d5f8b6715e4dc7b383eb3"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "der"
+version = "0.7.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e7c1832837b905bbfb5101e07cc24c8deddf52f93225eee6ead5f4d63d53ddcb"
+dependencies = [
+ "const-oid",
+ "zeroize",
+]
+
+[[package]]
+name = "digest"
+version = "0.10.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9ed9a281f7bc9b7576e61468ba615a66a5c8cfdff42420a70aa82701a3b1e292"
+dependencies = [
+ "block-buffer",
+ "crypto-common",
+]
+
+[[package]]
+name = "discord-channel"
+version = "0.2.0"
+dependencies = [
+ "ed25519-dalek",
+ "hex",
 "serde",
 "serde_json",
 "wit-bindgen",
 ]

+[[package]]
+name = "ed25519"
+version = "2.2.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "115531babc129696a58c64a4fef0a8bf9e9698629fb97e9e40767d235cfbcd53"
+dependencies = [
+ "pkcs8",
+ "signature",
+]
+
+[[package]]
+name = "ed25519-dalek"
+version = "2.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "70e796c081cee67dc755e1a36a0a172b897fab85fc3f6bc48307991f64e4eca9"
+dependencies = [
+ "curve25519-dalek",
+ "ed25519",
+ "serde",
+ "sha2",
+ "subtle",
+ "zeroize",
+]
+
 [[package]]
 name = "equivalent"
 version = "1.0.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "877a4ace8713b0bcf2a4e7eec82529c029f1d0619886d18145fea96c3ffe5c0f"

+[[package]]
+name = "fiat-crypto"
+version = "0.2.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "28dea519a9695b9977216879a3ebfddf92f1c08c05d984f8996aecd6ecdc811d"
+
+[[package]]
+name = "generic-array"
+version = "0.14.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "85649ca51fd72272d7821adaf274ad91c288277713d9c18820d8499a7ff69e9a"
+dependencies = [
+ "typenum",
+ "version_check",
+]
+
 [[package]]
 name = "hashbrown"
 version = "0.14.5"
@@ -68,6 +197,12 @@ version = "0.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2304e00983f87ffb38b55b444b5e3b60a884b5d30c0fca7d82fe33449bbe55ea"

+[[package]]
+name = "hex"
+version = "0.4.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7f24254aa9a54b5c858eaee2f5bccdb46aaf0e486a595ed5fd8f86ba55232a70"
+
 [[package]]
 name = "id-arena"
 version = "2.3.0"
@@ -98,6 +233,12 @@ version = "0.2.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "884e2677b40cc8c339eaefcb701c32ef1fd2493d71118dc0ca4b6a736c93bd67"

+[[package]]
+name = "libc"
+version = "0.2.182"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6800badb6cb2082ffd7b6a67e6125bb39f18782f793520caee8cb8846be06112"
+
 [[package]]
 name = "log"
 version = "0.4.29"
@@ -116,6 +257,16 @@ version = "1.21.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "42f5e15c9953c5e4ccceeb2e7382a716482c34515315f7b03532b8b4e8393d2d"

+[[package]]
+name = "pkcs8"
+version = "0.10.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f950b2377845cebe5cf8b5165cb3cc1a5e0fa5cfa3e1f7f55707d8fd82e0a7b7"
+dependencies = [
+ "der",
+ "spki",
+]
+
 [[package]]
 name = "prettyplease"
 version = "0.2.37"
@@ -144,6 +295,15 @@ dependencies = [
 "proc-macro2",
 ]

+[[package]]
+name = "rustc_version"
+version = "0.4.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cfcb3a22ef46e85b45de6ee7e79d063319ebb6594faafcf1c225ea92ab6e9b92"
+dependencies = [
+ "semver",
+]
+
 [[package]]
 name = "semver"
 version = "1.0.27"
@@ -193,6 +353,23 @@ dependencies = [
 "zmij",
 ]

+[[package]]
+name = "sha2"
+version = "0.10.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a7507d819769d01a365ab707794a4084392c824f54a7a6a7862f8c3d0892b283"
+dependencies = [
+ "cfg-if",
+ "cpufeatures",
+ "digest",
+]
+
+[[package]]
+name = "signature"
+version = "2.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "77549399552de45a898a580c1b41d445bf730df867cc44e6c0233bbc4b8329de"
+
 [[package]]
 name = "smallvec"
 version = "1.15.1"
@@ -208,6 +385,22 @@ dependencies = [
 "smallvec",
 ]

+[[package]]
+name = "spki"
+version = "0.7.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d91ed6c858b01f942cd56b37a94b3e0a1798290327d1236e4d9cf4eaca44d29d"
+dependencies = [
+ "base64ct",
+ "der",
+]
+
+[[package]]
+name = "subtle"
+version = "2.6.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "13c2bddecc57b384dee18652358fb23172facb8a2c51ccc10d74c157bdea3292"
+
 [[package]]
 name = "syn"
 version = "2.0.117"
@@ -219,6 +412,12 @@ dependencies = [
 "unicode-ident",
 ]

+[[package]]
+name = "typenum"
+version = "1.19.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "562d481066bde0658276a35467c4af00bdc6ee726305698a55b86e61d7ad82bb"
+
 [[package]]
 name = "unicode-ident"
 version = "1.0.24"
@@ -394,6 +593,12 @@ dependencies = [
 "syn",
 ]

+[[package]]
+name = "zeroize"
+version = "1.8.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b97154e67e32c85465826e8bcc1c59429aaaf107c1e4a9e53c8d8ccd5eff88d0"
+
 [[package]]
 name = "zmij"
 version = "1.0.21"
@@ -10,6 +10,8 @@ publish = false
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 wit-bindgen = "0.36"
+ed25519-dalek = { version = "2", default-features = false, features = ["alloc", "fast", "zeroize"] }
+hex = "0.4"

 [lib]
 crate-type = ["cdylib"]
@@ -21,11 +21,10 @@ WASM channel for Discord integration - handle slash commands and button interact
   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.
+   **Note:** The `discord_bot_token` secret is used for Discord REST API calls.
+   Interaction signature verification is performed inside the Discord channel
+   module and uses the channel config field `webhook_secret` (set this to your
+   Discord app public key hex).

 ## Discord Configuration

@@ -87,6 +86,30 @@ If an internal error occurs (e.g., metadata serialization failure), the tool att
 Check the host logs for detailed error information.

 ## Advanced Usage
+### Mention Polling
+
+The Discord channel can also poll configured channels for `@bot` mentions.
+
+Example channel config:
+
+```json
+{
+  "require_signature_verification": true,
+  "webhook_secret": "YOUR_DISCORD_PUBLIC_KEY_HEX",
+  "polling_enabled": true,
+  "poll_interval_ms": 30000,
+  "mention_channel_ids": ["123456789012345678"],
+  "owner_id": null,
+  "dm_policy": "pairing",
+  "allow_from": []
+}
+```
+
+### Access Control
+
+- `owner_id`: when set, only that Discord user can interact with the bot.
+- `dm_policy`: `open` allows all DMs; `pairing` requires approval.
+- `allow_from`: allowlist entries for DM pairing checks (`*`, user id, or username).

 ### Embeds

@@ -96,8 +119,11 @@ To send embeds, include an `embeds` array in the `metadata_json` field of the ag

 ### "Invalid Signature"

- Check that `discord_public_key` is set correctly in IronClaw secrets.
- This validation happens on the host before reaching the WASM.
+- Check that `webhook_secret` is set to your Discord app public key hex in the
+  Discord channel config.
+- Validation happens inside the Discord WASM channel.
+- If `require_signature_verification` is `true` and `webhook_secret` is empty,
+  the channel returns HTTP `500` with a configuration error.

 ### "401 Unauthorized"

@@ -3,7 +3,7 @@
  "wit_version": "0.3.0",
  "type": "channel",
  "name": "discord",
-  "description": "Discord Gateway/Webhook channel for handling slash commands, buttons, and messages",
+  "description": "Discord webhook channel for slash commands, components, and optional mention polling",
  "setup": {
    "required_secrets": [
      {
@@ -41,7 +41,7 @@
    },
    "channel": {
      "allowed_paths": ["/webhook/discord"],
-      "allow_polling": false,
+      "allow_polling": true,
      "callback_timeout_secs": 45,
      "workspace_prefix": "channels/discord/",
      "emit_rate_limit": {
@@ -55,8 +55,12 @@
  },
  "config": {
    "require_signature_verification": true,
+    "webhook_secret": null,
+    "polling_enabled": false,
+    "poll_interval_ms": 30000,
+    "mention_channel_ids": [],
    "owner_id": null,
    "dm_policy": "pairing",
    "allow_from": []
  }
-}
+}
@@ -1,5 +1,10 @@
 # WARNING: Replace all CHANGE_ME values before deploying.
 # Do not use placeholder passwords in production.
+
+# Pin the Docker image version for deterministic deployments.
+# Update this value when deploying a new release.
+# IRONCLAW_VERSION=v1.0.0
+
 DATABASE_URL=postgres://ironclaw:CHANGE_ME@localhost:5432/ironclaw

 # NEAR AI Cloud (API key auth, Chat Completions API)
@@ -5,13 +5,17 @@ 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 \
+EnvironmentFile=/opt/ironclaw/.env
+# Pin to a specific version tag or digest instead of :latest to prevent
+# uncontrolled deployments. Update IRONCLAW_VERSION in /opt/ironclaw/.env
+# or replace the tag below when deploying a new release.
+ExecStartPre=/bin/bash -c 'docker pull us-central1-docker.pkg.dev/ironclaw-prod/ironclaw/agent:${IRONCLAW_VERSION:-latest}'
+ExecStart=/bin/bash -c 'docker run --rm \
  --name ironclaw \
  --env-file /opt/ironclaw/.env \
-  --network=host \
-  us-central1-docker.pkg.dev/ironclaw-prod/ironclaw/agent:latest \
-  --no-onboard
+  -p 3000:3000 \
+  us-central1-docker.pkg.dev/ironclaw-prod/ironclaw/agent:${IRONCLAW_VERSION:-latest} \
+  --no-onboard'
 ExecStop=/usr/bin/docker stop ironclaw
 Restart=always
 RestartSec=10
@@ -24,8 +24,15 @@ systemctl enable docker
 systemctl start docker

 echo "==> Installing Cloud SQL Auth Proxy"
+CLOUD_SQL_PROXY_VERSION="v2.14.3"
+CLOUD_SQL_PROXY_SHA256="75e7cc1f158ab6f97b7810e9d8419c55735cff40bc56d4f19673adfdf2406a59"
 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
+  "https://storage.googleapis.com/cloud-sql-connectors/cloud-sql-proxy/${CLOUD_SQL_PROXY_VERSION}/cloud-sql-proxy.linux.amd64"
+echo "${CLOUD_SQL_PROXY_SHA256}  /usr/local/bin/cloud-sql-proxy" | sha256sum -c - || {
+  echo "ERROR: Cloud SQL Auth Proxy checksum verification failed -- aborting"
+  rm -f /usr/local/bin/cloud-sql-proxy
+  exit 1
+}
 chmod +x /usr/local/bin/cloud-sql-proxy

 echo "==> Installing systemd services"
@@ -15,6 +15,7 @@ configurations.
 | io.net | `ionet` | `IONET_API_KEY` | Intelligence API |
 | Mistral | `mistral` | `MISTRAL_API_KEY` | Mistral models |
 | Yandex AI Studio | `yandex` | `YANDEX_API_KEY` | YandexGPT models |
+| MiniMax | `minimax` | `MINIMAX_API_KEY` | MiniMax-M2.5 models |
 | Cloudflare Workers AI | `cloudflare` | `CLOUDFLARE_API_KEY` | Access to Workers AI |
 | Ollama | `ollama` | No | Local inference |
 | AWS Bedrock | `bedrock` | AWS credentials | Native Converse API |
@@ -74,6 +75,25 @@ Pull a model first: `ollama pull llama3.2`

 ---

+## MiniMax
+
+[MiniMax](https://platform.minimax.io) provides high-performance language models with 204,800 token context windows.
+
+```env
+LLM_BACKEND=minimax
+MINIMAX_API_KEY=...
+```
+
+Available models: `MiniMax-M2.5` (default), `MiniMax-M2.5-highspeed`
+
+To use the China mainland endpoint, set:
+
+```env
+MINIMAX_BASE_URL=https://api.minimaxi.com/v1
+```
+
+---
+
 ## AWS Bedrock (requires `--features bedrock`)

 Uses the native AWS Converse API via `aws-sdk-bedrockruntime`. Supports standard AWS
@@ -382,6 +382,27 @@
      "can_list_models": true
    }
  },
+  {
+    "id": "minimax",
+    "aliases": [
+      "mini_max"
+    ],
+    "protocol": "open_ai_completions",
+    "default_base_url": "https://api.minimax.io/v1",
+    "api_key_env": "MINIMAX_API_KEY",
+    "api_key_required": true,
+    "base_url_env": "MINIMAX_BASE_URL",
+    "model_env": "MINIMAX_MODEL",
+    "default_model": "MiniMax-M2.5",
+    "description": "MiniMax API (MiniMax-M2.5 and MiniMax-M2.5-highspeed models)",
+    "setup": {
+      "kind": "api_key",
+      "secret_name": "llm_minimax_api_key",
+      "key_url": "https://platform.minimax.io",
+      "display_name": "MiniMax",
+      "can_list_models": false
+    }
+  },
  {
    "id": "cloudflare",
    "aliases": [
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/discord-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/discord-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "efa1b9019fa33e243f8db1e1fcc732731d45836336bdd26ca19b6fe227ca8b69"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/slack-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/slack-0.2.1-wasm32-wasip2.tar.gz",
+      "sha256": "d4667e35126986509d862bc3a0088777305d8f41c75de83c1e223b42312ede48"
    }
  },
  "auth_summary": {
@@ -2,7 +2,7 @@
  "name": "telegram",
  "display_name": "Telegram Channel",
  "kind": "channel",
-  "version": "0.2.2",
+  "version": "0.2.3",
  "wit_version": "0.3.0",
  "description": "Talk to your agent through a Telegram bot",
  "keywords": [
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/telegram-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/telegram-0.2.3-wasm32-wasip2.tar.gz",
+      "sha256": "b9a83d5a2d1285ce0ec116b354336a1f245f893291ccb01dffbcaccf89d72aed"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/whatsapp-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/whatsapp-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "feb9194719d9bed796b070ab4dc30348dbfb5d3dec56f9f21e02d14137abab01"
    }
  },
  "auth_summary": {
@@ -2,7 +2,7 @@
  "name": "github",
  "display_name": "GitHub",
  "kind": "tool",
-  "version": "0.2.1",
+  "version": "0.2.0",
  "wit_version": "0.3.0",
  "description": "GitHub integration for issues, PRs, repos, and code search",
  "keywords": [
@@ -19,8 +19,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/github-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/github-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "da9fac56b6f20197a415489bbaec9fefb085a5cf6324cab79ea48a47eb19c13b"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/gmail-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/gmail-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "ee9574e02e92bc1d481f1310eb88afd99ee52bf6971074ab33bd76bf99b34b1d"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/google-calendar-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/google-calendar-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "2fa47150ea222e787c122182ad6f4dfa2ffaf5fe490d05e8de887a76445f8d2d"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/google-docs-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/google-docs-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "40e134a1c1564f832ca861c3396895d4e33ec67b99313fc1f97baf8d971423a9"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/google-drive-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/google-drive-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "002a341a1d58125563a7c69561b26fbc2629b04ea723cade744102bdc0fbb71f"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/google-sheets-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/google-sheets-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "8aa2c9d52f033edea3a6c2311b0ec694ccb6d0a54ef07e94d72bf8be1ce8009a"
    }
  },
  "auth_summary": {
@@ -17,8 +17,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/google-slides-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/google-slides-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "e931a97d4fd0b0b938e464dc7c7f2be6ea6b4d1508f5ea3cd931d44db23f05f5"
    }
  },
  "auth_summary": {
@@ -0,0 +1,41 @@
+{
+  "name": "llm-context",
+  "display_name": "LLM Context",
+  "kind": "tool",
+  "version": "0.1.0",
+  "wit_version": "0.3.0",
+  "description": "Fetch pre-extracted web content from Brave Search for grounding LLM answers (RAG, fact-checking)",
+  "keywords": [
+    "search",
+    "web",
+    "brave",
+    "rag",
+    "grounding",
+    "llm",
+    "context"
+  ],
+  "source": {
+    "dir": "tools-src/llm-context",
+    "capabilities": "llm-context-tool.capabilities.json",
+    "crate_name": "llm-context-tool"
+  },
+  "artifacts": {
+    "wasm32-wasip2": {
+      "url": "https://github.com/nearai/ironclaw/releases/latest/download/llm-context-wasm32-wasip2.tar.gz",
+      "sha256": "581cc5867ef3b75116b7ddc8161e63dd92befe2b53e6ad8213c007639aa243c3"
+    }
+  },
+  "auth_summary": {
+    "method": "manual",
+    "provider": "Brave",
+    "secrets": [
+      "brave_api_key"
+    ],
+    "shared_auth": "Same API key as Web Search tool (brave_api_key)",
+    "setup_url": "https://brave.com/search/api/"
+  },
+  "tags": [
+    "default",
+    "search"
+  ]
+}
@@ -17,8 +17,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/slack-tool-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/slack-0.2.1-wasm32-wasip2.tar.gz",
+      "sha256": "d4667e35126986509d862bc3a0088777305d8f41c75de83c1e223b42312ede48"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/telegram-mtproto-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/telegram-0.2.2-wasm32-wasip2.tar.gz",
+      "sha256": "b9a83d5a2d1285ce0ec116b354336a1f245f893291ccb01dffbcaccf89d72aed"
    }
  },
  "auth_summary": {
@@ -18,8 +18,8 @@
  },
  "artifacts": {
    "wasm32-wasip2": {
-      "url": "https://github.com/nearai/ironclaw/releases/latest/download/web-search-wasm32-wasip2.tar.gz",
-      "sha256": null
+      "url": "https://github.com/nearai/ironclaw/releases/download/v0.18.0/web-search-0.2.0-wasm32-wasip2.tar.gz",
+      "sha256": "56834573c54ea2a33cea1eb0f04bbdf59f1ef8d8702995cf431b0921302eeccc"
    }
  },
  "auth_summary": {
@@ -18,7 +18,7 @@ use crate::agent::self_repair::{DefaultSelfRepair, RepairResult, SelfRepair};
 use crate::agent::session_manager::SessionManager;
 use crate::agent::submission::{Submission, SubmissionParser, SubmissionResult};
 use crate::agent::{HeartbeatConfig as AgentHeartbeatConfig, Router, Scheduler};
-use crate::channels::{ChannelManager, IncomingMessage, OutgoingResponse, StatusUpdate};
+use crate::channels::{ChannelManager, IncomingMessage, OutgoingResponse};
 use crate::config::{AgentConfig, HeartbeatConfig, RoutineConfig, SkillsConfig};
 use crate::context::ContextManager;
 use crate::db::Database;
@@ -936,29 +936,10 @@ impl Agent {
            SubmissionResult::Ok { message } => Ok(message),
            SubmissionResult::Error { message } => Ok(Some(format!("Error: {}", message))),
            SubmissionResult::Interrupted => Ok(Some("Interrupted.".into())),
-            SubmissionResult::NeedApproval {
-                request_id,
-                tool_name,
-                description,
-                parameters,
-            } => {
-                // Each channel renders the approval prompt via send_status.
-                // Web gateway shows an inline card, REPL prints a formatted prompt, etc.
-                let _ = self
-                    .channels
-                    .send_status(
-                        &message.channel,
-                        StatusUpdate::ApprovalNeeded {
-                            request_id: request_id.to_string(),
-                            tool_name,
-                            description,
-                            parameters,
-                        },
-                        &message.metadata,
-                    )
-                    .await;
-
-                // Empty string signals the caller to skip respond() (no duplicate text)
+            SubmissionResult::NeedApproval { .. } => {
+                // ApprovalNeeded status was already sent by thread_ops.rs before
+                // returning this result. Empty string signals the caller to skip
+                // respond() (no duplicate text).
                Ok(Some(String::new()))
            }
        }
@@ -554,6 +554,31 @@ impl<'a> LoopDelegate for ChatDelegate<'a> {
                };

                if needs_approval {
+                    // In non-DM relay channels, auto-deny approval-
+                    // requiring tools to prevent stuck AwaitingApproval
+                    // state and prompt injection from other users.
+                    let is_relay = self.message.channel.ends_with("-relay");
+                    let is_dm = self
+                        .message
+                        .metadata
+                        .get("event_type")
+                        .and_then(|v| v.as_str())
+                        == Some("direct_message");
+                    if is_relay && !is_dm {
+                        tracing::info!(
+                            tool = %tc.name,
+                            channel = %self.message.channel,
+                            "Auto-denying approval-requiring tool in non-DM relay channel"
+                        );
+                        let reject_msg = format!(
+                            "Tool '{}' requires approval and cannot run in shared channels. \
+                             Ask the user to message me directly (DM) to use this tool.",
+                            tc.name
+                        );
+                        preflight.push((tc, PreflightOutcome::Rejected(reject_msg)));
+                        continue;
+                    }
+
                    approval_needed = Some((idx, tc, tool));
                    break;
                }
@@ -2235,4 +2260,51 @@ mod tests {
            "Present 'data' field should produce non-empty string"
        );
    }
+
+    /// Test the relay channel auto-deny decision logic:
+    /// approval-requiring tools in non-DM relay channels must be rejected.
+    #[test]
+    fn test_relay_non_dm_auto_deny_decision() {
+        use crate::channels::IncomingMessage;
+
+        // Case 1: relay channel + non-DM → should auto-deny
+        let msg = IncomingMessage::new("slack-relay", "u1", "hello")
+            .with_metadata(serde_json::json!({ "event_type": "message" }));
+        let is_relay = msg.channel.ends_with("-relay");
+        let is_dm =
+            msg.metadata.get("event_type").and_then(|v| v.as_str()) == Some("direct_message");
+        assert!(is_relay && !is_dm, "Should auto-deny in relay non-DM");
+
+        // Case 2: relay channel + DM → should NOT auto-deny
+        let msg_dm = IncomingMessage::new("slack-relay", "u1", "hello")
+            .with_metadata(serde_json::json!({ "event_type": "direct_message" }));
+        let is_dm_2 =
+            msg_dm.metadata.get("event_type").and_then(|v| v.as_str()) == Some("direct_message");
+        assert!(
+            !msg_dm.channel.ends_with("-relay") || is_dm_2,
+            "Should NOT auto-deny in relay DM"
+        );
+
+        // Case 3: non-relay channel → should NOT auto-deny
+        let msg_web = IncomingMessage::new("web", "u1", "hello")
+            .with_metadata(serde_json::json!({ "event_type": "message" }));
+        assert!(
+            !msg_web.channel.ends_with("-relay"),
+            "Non-relay channel should not trigger auto-deny"
+        );
+    }
+
+    /// Test that the auto-deny produces a PreflightOutcome::Rejected-style message.
+    #[test]
+    fn test_relay_auto_deny_message_format() {
+        let tool_name = "shell";
+        let result_msg = format!(
+            "Tool '{}' requires approval and cannot run in shared channels. \
+             Ask the user to message me directly (DM) to use this tool.",
+            tool_name
+        );
+        assert!(result_msg.contains("shell"));
+        assert!(result_msg.contains("approval"));
+        assert!(result_msg.contains("DM"));
+    }
 }
@@ -32,7 +32,6 @@ pub mod task;
 mod thread_ops;
 pub mod undo;

-pub use crate::worker::{Worker, WorkerDeps};
 pub(crate) use agent_loop::truncate_for_preview;
 pub use agent_loop::{Agent, AgentDeps};
 pub use compaction::{CompactionResult, ContextCompactor};
@@ -207,7 +207,7 @@ impl Trigger {
 #[derive(Debug, Clone, Serialize, Deserialize)]
 #[serde(tag = "type", rename_all = "snake_case")]
 pub enum RoutineAction {
-    /// Single LLM call, no tools. Cheap and fast.
+    /// Single LLM call (optionally with tools). Cheap and fast.
    Lightweight {
        /// The prompt sent to the LLM.
        prompt: String,
@@ -217,6 +217,14 @@ pub enum RoutineAction {
        /// Max output tokens (default: 4096).
        #[serde(default = "default_max_tokens")]
        max_tokens: u32,
+        /// Enable tool access (default: false for backward compatibility).
+        /// When true, the LLM can call tools during execution.
+        /// Tools requiring approval are automatically filtered out.
+        #[serde(default)]
+        use_tools: bool,
+        /// Max tool call rounds (default: 3). Only used when use_tools is true.
+        #[serde(default = "default_max_tool_rounds")]
+        max_tool_rounds: u32,
    },
    /// Full multi-turn worker job with tool access.
    FullJob {
@@ -243,6 +251,19 @@ fn default_max_iterations() -> u32 {
    10
 }

+fn default_max_tool_rounds() -> u32 {
+    3
+}
+
+/// Hard upper bound for max_tool_rounds to prevent runaway loops and cost explosion.
+pub(crate) const MAX_TOOL_ROUNDS_LIMIT: u32 = 20;
+
+/// Clamp max_tool_rounds to [1, MAX_TOOL_ROUNDS_LIMIT].
+/// Accepts u64 to avoid truncation before clamping.
+fn clamp_max_tool_rounds(value: u64) -> u32 {
+    value.clamp(1, MAX_TOOL_ROUNDS_LIMIT as u64) as u32
+}
+
 /// Parse a `tool_permissions` JSON array into a `Vec<String>`.
 pub fn parse_tool_permissions(value: &serde_json::Value) -> Vec<String> {
    value
@@ -290,10 +311,22 @@ impl RoutineAction {
                    .get("max_tokens")
                    .and_then(|v| v.as_u64())
                    .unwrap_or(default_max_tokens() as u64) as u32;
+                let use_tools = config
+                    .get("use_tools")
+                    .and_then(|v| v.as_bool())
+                    .unwrap_or(false);
+                let max_tool_rounds = clamp_max_tool_rounds(
+                    config
+                        .get("max_tool_rounds")
+                        .and_then(|v| v.as_u64())
+                        .unwrap_or(default_max_tool_rounds() as u64),
+                );
                Ok(RoutineAction::Lightweight {
                    prompt,
                    context_paths,
                    max_tokens,
+                    use_tools,
+                    max_tool_rounds,
                })
            }
            "full_job" => {
@@ -339,10 +372,14 @@ impl RoutineAction {
                prompt,
                context_paths,
                max_tokens,
+                use_tools,
+                max_tool_rounds,
            } => serde_json::json!({
                "prompt": prompt,
                "context_paths": context_paths,
                "max_tokens": max_tokens,
+                "use_tools": use_tools,
+                "max_tool_rounds": max_tool_rounds,
            }),
            RoutineAction::FullJob {
                title,
@@ -504,7 +541,8 @@ pub fn next_cron_fire(
 #[cfg(test)]
 mod tests {
    use crate::agent::routine::{
-        RoutineAction, RoutineGuardrails, RunStatus, Trigger, content_hash, next_cron_fire,
+        MAX_TOOL_ROUNDS_LIMIT, RoutineAction, RoutineGuardrails, RunStatus, Trigger, content_hash,
+        next_cron_fire,
    };

    #[test]
@@ -554,11 +592,13 @@ mod tests {
            prompt: "Check PRs".to_string(),
            context_paths: vec!["context/priorities.md".to_string()],
            max_tokens: 2048,
+            use_tools: false,
+            max_tool_rounds: 3,
        };
        let json = action.to_config_json();
        let parsed = RoutineAction::from_db("lightweight", json).expect("parse lightweight");
        assert!(
-            matches!(parsed, RoutineAction::Lightweight { prompt, context_paths, max_tokens }
+            matches!(parsed, RoutineAction::Lightweight { prompt, context_paths, max_tokens, .. }
            if prompt == "Check PRs" && context_paths.len() == 1 && max_tokens == 2048)
        );
    }
@@ -695,4 +735,77 @@ mod tests {
        );
        assert_eq!(Trigger::Manual.type_tag(), "manual");
    }
+
+    #[test]
+    fn test_action_lightweight_backward_compat_no_use_tools() {
+        // Simulate old DB record without use_tools field
+        let json = serde_json::json!({
+            "prompt": "old routine",
+            "context_paths": [],
+            "max_tokens": 4096
+        });
+        let parsed = RoutineAction::from_db("lightweight", json).expect("parse lightweight");
+        assert!(
+            matches!(parsed, RoutineAction::Lightweight { use_tools, max_tool_rounds, .. }
+            if !use_tools && max_tool_rounds == 3),
+            "missing use_tools should default to false, max_tool_rounds to 3"
+        );
+    }
+
+    #[test]
+    fn test_max_tool_rounds_clamped_to_upper_bound() {
+        let json = serde_json::json!({
+            "prompt": "test",
+            "use_tools": true,
+            "max_tool_rounds": 9999
+        });
+        let parsed = RoutineAction::from_db("lightweight", json).expect("parse");
+        match parsed {
+            RoutineAction::Lightweight {
+                max_tool_rounds, ..
+            } => {
+                assert_eq!(
+                    max_tool_rounds, MAX_TOOL_ROUNDS_LIMIT,
+                    "should clamp to MAX_TOOL_ROUNDS_LIMIT"
+                );
+            }
+            _ => panic!("expected Lightweight"),
+        }
+    }
+
+    #[test]
+    fn test_max_tool_rounds_clamped_to_lower_bound() {
+        let json = serde_json::json!({
+            "prompt": "test",
+            "use_tools": true,
+            "max_tool_rounds": 0
+        });
+        let parsed = RoutineAction::from_db("lightweight", json).expect("parse");
+        match parsed {
+            RoutineAction::Lightweight {
+                max_tool_rounds, ..
+            } => {
+                assert_eq!(max_tool_rounds, 1, "should clamp 0 to 1");
+            }
+            _ => panic!("expected Lightweight"),
+        }
+    }
+
+    #[test]
+    fn test_max_tool_rounds_normal_value_passes_through() {
+        let json = serde_json::json!({
+            "prompt": "test",
+            "use_tools": true,
+            "max_tool_rounds": 10
+        });
+        let parsed = RoutineAction::from_db("lightweight", json).expect("parse");
+        match parsed {
+            RoutineAction::Lightweight {
+                max_tool_rounds, ..
+            } => {
+                assert_eq!(max_tool_rounds, 10, "normal value should pass through");
+            }
+            _ => panic!("expected Lightweight"),
+        }
+    }
 }
@@ -459,7 +459,20 @@ async fn execute_routine(ctx: EngineContext, routine: Routine, run: RoutineRun)
            prompt,
            context_paths,
            max_tokens,
-        } => execute_lightweight(&ctx, &routine, prompt, context_paths, *max_tokens).await,
+            use_tools,
+            max_tool_rounds,
+        } => {
+            execute_lightweight(
+                &ctx,
+                &routine,
+                prompt,
+                context_paths,
+                *max_tokens,
+                *use_tools,
+                *max_tool_rounds,
+            )
+            .await
+        }
        RoutineAction::FullJob {
            title,
            description,
@@ -670,6 +683,8 @@ async fn execute_lightweight(
    prompt: &str,
    context_paths: &[String],
    max_tokens: u32,
+    use_tools: bool,
+    max_tool_rounds: u32,
 ) -> Result<(RunStatus, Option<String>, Option<i32>), RoutineError> {
    // Load context from workspace
    let mut context_parts = Vec::new();
@@ -732,14 +747,15 @@ async fn execute_lightweight(
        Err(_) => max_tokens,
    };

-    // If tools are enabled, use the tool execution loop; otherwise, single LLM call
-    if ctx.config.lightweight_tools_enabled {
+    // If tools are enabled (both globally and per-routine), use the tool execution loop
+    if use_tools && ctx.config.lightweight_tools_enabled {
        execute_lightweight_with_tools(
            ctx,
            routine,
            &system_prompt,
            &full_prompt,
            effective_max_tokens,
+            max_tool_rounds,
        )
        .await
    } else {
@@ -783,24 +799,12 @@ async fn execute_lightweight_no_tools(
            reason: e.to_string(),
        })?;

-    let content = response.content.trim();
-    let tokens_used = Some((response.input_tokens + response.output_tokens) as i32);
-
-    // Empty content guard
-    if content.is_empty() {
-        return if response.finish_reason == FinishReason::Length {
-            Err(RoutineError::TruncatedResponse)
-        } else {
-            Err(RoutineError::EmptyResponse)
-        };
-    }
-
-    // Check for the "nothing to do" sentinel
-    if content == "ROUTINE_OK" || content.contains("ROUTINE_OK") {
-        return Ok((RunStatus::Ok, None, tokens_used));
-    }
-
-    Ok((RunStatus::Attention, Some(content.to_string()), tokens_used))
+    handle_text_response(
+        &response.content,
+        response.finish_reason,
+        response.input_tokens,
+        response.output_tokens,
+    )
 }

 /// Handle a text-only LLM response in lightweight routine execution.
@@ -850,6 +854,7 @@ async fn execute_lightweight_with_tools(
    system_prompt: &str,
    full_prompt: &str,
    effective_max_tokens: u32,
+    max_tool_rounds: u32,
 ) -> Result<(RunStatus, Option<String>, Option<i32>), RoutineError> {
    let mut messages = if system_prompt.is_empty() {
        vec![ChatMessage::user(full_prompt)]
@@ -860,7 +865,9 @@ async fn execute_lightweight_with_tools(
        ]
    };

-    let max_iterations = ctx.config.lightweight_max_iterations.min(5);
+    let max_iterations = max_tool_rounds
+        .min(ctx.config.lightweight_max_iterations)
+        .min(5);
    let mut iteration = 0;
    let mut total_input_tokens = 0;
    let mut total_output_tokens = 0;
@@ -906,7 +913,10 @@ async fn execute_lightweight_with_tools(
            );
        } else {
            // Tool-enabled iteration
-            let tool_defs = ctx.tools.tool_definitions().await;
+            let tool_defs = ctx
+                .tools
+                .tool_definitions_excluding(ROUTINE_TOOL_DENYLIST)
+                .await;

            let request = ToolCompletionRequest::new(messages.clone(), tool_defs)
                .with_max_tokens(effective_max_tokens)
@@ -972,12 +982,33 @@ async fn execute_lightweight_with_tools(
    }
 }

+/// Tools that must never be callable from lightweight routines.
+///
+/// These tools pose autonomy-escalation risks: a routine could self-replicate,
+/// modify its own triggers/prompts, delete other routines, or restart the agent.
+const ROUTINE_TOOL_DENYLIST: &[&str] = &[
+    "routine_create",
+    "routine_update",
+    "routine_delete",
+    "routine_fire",
+    "restart",
+];
+
 /// Execute a single tool for a lightweight routine.
 async fn execute_routine_tool(
    ctx: &EngineContext,
    job_ctx: &JobContext,
    tc: &ToolCall,
 ) -> Result<String, Box<dyn std::error::Error + Send + Sync>> {
+    // Block tools that pose autonomy-escalation risks
+    if ROUTINE_TOOL_DENYLIST.contains(&tc.name.as_str()) {
+        return Err(format!(
+            "Tool '{}' is not available in lightweight routines",
+            tc.name
+        )
+        .into());
+    }
+
    // Check if tool exists
    let tool = ctx
        .tools
@@ -1119,9 +1150,11 @@ pub fn spawn_cron_ticker(
    interval: Duration,
 ) -> tokio::task::JoinHandle<()> {
    tokio::spawn(async move {
+        // Run one check immediately so routines due at startup don't wait
+        // an extra full polling interval.
+        engine.check_cron_triggers().await;
+
        let mut ticker = tokio::time::interval(interval);
-        // Skip immediate first tick
-        ticker.tick().await;

        loop {
            ticker.tick().await;
@@ -1283,6 +1316,36 @@ mod tests {
        }
    }

+    #[test]
+    fn test_routine_tool_denylist_blocks_self_management_tools() {
+        let denylisted = vec![
+            "routine_create",
+            "routine_update",
+            "routine_delete",
+            "routine_fire",
+            "restart",
+        ];
+        for tool in &denylisted {
+            assert!(
+                super::ROUTINE_TOOL_DENYLIST.contains(tool),
+                "Tool '{}' should be in ROUTINE_TOOL_DENYLIST",
+                tool
+            );
+        }
+    }
+
+    #[test]
+    fn test_routine_tool_denylist_allows_safe_tools() {
+        let allowed = vec!["echo", "time", "json", "http", "memory_search", "shell"];
+        for tool in &allowed {
+            assert!(
+                !super::ROUTINE_TOOL_DENYLIST.contains(tool),
+                "Tool '{}' should NOT be in ROUTINE_TOOL_DENYLIST",
+                tool
+            );
+        }
+    }
+
    #[test]
    fn test_empty_response_handling() {
        // Simulate the empty content guard logic
@@ -1297,4 +1360,11 @@ mod tests {
        assert_eq!(finish_reason_length, crate::llm::FinishReason::Length);
        assert_eq!(finish_reason_stop, crate::llm::FinishReason::Stop);
    }
+
+    #[test]
+    fn test_truncate_adds_ellipsis_when_over_limit() {
+        let input = "abcdefghijk";
+        let out = super::truncate(input, 5);
+        assert_eq!(out, "abcde...");
+    }
 }
@@ -486,7 +486,12 @@ impl Agent {
                    .channels
                    .send_status(
                        &message.channel,
-                        StatusUpdate::Status("Awaiting approval".into()),
+                        StatusUpdate::ApprovalNeeded {
+                            request_id: request_id.to_string(),
+                            tool_name: tool_name.clone(),
+                            description: description.clone(),
+                            parameters: parameters.clone(),
+                        },
                        &message.metadata,
                    )
                    .await;
@@ -1297,7 +1302,12 @@ impl Agent {
                    .channels
                    .send_status(
                        &message.channel,
-                        StatusUpdate::Status("Awaiting approval".into()),
+                        StatusUpdate::ApprovalNeeded {
+                            request_id: request_id.to_string(),
+                            tool_name: tool_name.clone(),
+                            description: description.clone(),
+                            parameters: parameters.clone(),
+                        },
                        &message.metadata,
                    )
                    .await;
@@ -1368,7 +1378,12 @@ impl Agent {
                        .channels
                        .send_status(
                            &message.channel,
-                            StatusUpdate::Status("Awaiting approval".into()),
+                            StatusUpdate::ApprovalNeeded {
+                                request_id: request_id.to_string(),
+                                tool_name: tool_name.clone(),
+                                description: description.clone(),
+                                parameters: parameters.clone(),
+                            },
                            &message.metadata,
                        )
                        .await;
@@ -9,6 +9,7 @@

 use std::sync::Arc;

+use crate::agent::SessionManager as AgentSessionManager;
 use crate::channels::web::log_layer::LogBroadcaster;
 use crate::config::Config;
 use crate::context::ContextManager;
@@ -46,6 +47,8 @@ pub struct AppComponents {
    pub log_broadcaster: Arc<LogBroadcaster>,
    pub context_manager: Arc<ContextManager>,
    pub hooks: Arc<HookRegistry>,
+    /// Shared thread/session manager used by the standard agent runtime.
+    pub agent_session_manager: Arc<AgentSessionManager>,
    pub skill_registry: Option<Arc<std::sync::RwLock<SkillRegistry>>>,
    pub skill_catalog: Option<Arc<SkillCatalog>>,
    pub cost_guard: Arc<crate::agent::cost_guard::CostGuard>,
@@ -287,6 +290,7 @@ impl AppBuilder {
            Arc::new(ToolRegistry::new())
        };
        tools.register_builtin_tools();
+        tools.register_tool_info();

        if let Some(ref ss) = self.secrets_store {
            tools.register_secrets_tools(Arc::clone(ss));
@@ -300,7 +304,8 @@ impl AppBuilder {

        // Register memory tools if database is available
        let workspace = if let Some(ref db) = self.db {
-            let mut ws = Workspace::new_with_db("default", db.clone());
+            let mut ws = Workspace::new_with_db("default", db.clone())
+                .with_search_config(&self.config.search);
            if let Some(ref emb) = embeddings {
                ws = ws.with_embeddings(emb.clone());
            }
@@ -689,6 +694,8 @@ impl AppBuilder {

        // Create hook registry early so runtime extension activation can register hooks.
        let hooks = Arc::new(HookRegistry::new());
+        let agent_session_manager =
+            Arc::new(AgentSessionManager::new().with_hooks(Arc::clone(&hooks)));

        let (
            mcp_session_manager,
@@ -795,6 +802,7 @@ impl AppBuilder {
            log_broadcaster: self.log_broadcaster,
            context_manager,
            hooks,
+            agent_session_manager,
            skill_registry,
            skill_catalog,
            cost_guard,
@@ -805,3 +813,69 @@ impl AppBuilder {
        })
    }
 }
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use async_trait::async_trait;
+    use tokio::sync::mpsc;
+
+    use crate::agent::SessionManager as AgentSessionManager;
+    use crate::hooks::{
+        Hook, HookContext, HookError, HookEvent, HookOutcome, HookPoint, HookRegistry,
+    };
+
+    struct SessionStartHook {
+        tx: mpsc::UnboundedSender<(String, String)>,
+    }
+
+    #[async_trait]
+    impl Hook for SessionStartHook {
+        fn name(&self) -> &str {
+            "session-start-test"
+        }
+
+        fn hook_points(&self) -> &[HookPoint] {
+            &[HookPoint::OnSessionStart]
+        }
+
+        async fn execute(
+            &self,
+            event: &HookEvent,
+            _ctx: &HookContext,
+        ) -> Result<HookOutcome, HookError> {
+            if let HookEvent::SessionStart {
+                user_id,
+                session_id,
+            } = event
+            {
+                self.tx
+                    .send((user_id.clone(), session_id.clone()))
+                    .expect("test channel receiver should be alive");
+            } else {
+                panic!("SessionStartHook received an unexpected event: {event:?}");
+            }
+            Ok(HookOutcome::ok())
+        }
+    }
+
+    #[tokio::test]
+    async fn agent_session_manager_runs_session_start_hooks() {
+        let hooks = Arc::new(HookRegistry::new());
+        let (tx, mut rx) = mpsc::unbounded_channel();
+        hooks.register(Arc::new(SessionStartHook { tx })).await;
+
+        let manager = AgentSessionManager::new().with_hooks(Arc::clone(&hooks));
+        manager.get_or_create_session("user-123").await;
+
+        let (user_id, session_id) =
+            tokio::time::timeout(std::time::Duration::from_secs(1), rx.recv())
+                .await
+                .expect("session start hook should fire")
+                .expect("session start payload should be present");
+
+        assert_eq!(user_id, "user-123");
+        assert!(!session_id.is_empty());
+    }
+}
@@ -269,95 +269,105 @@ async fn webhook_handler(
    let mut fallback_req = None;
    {
        let webhook_secret = state.webhook_secret.read().await;
-        if let Some(expected_secret) = webhook_secret.as_ref() {
-            let expected_secret = expected_secret.expose_secret();
+        let Some(expected_secret) = webhook_secret.as_ref() else {
+            return (
+                StatusCode::UNAUTHORIZED,
+                Json(WebhookResponse {
+                    message_id: Uuid::nil(),
+                    status: "error".to_string(),
+                    response: Some(
+                        "Webhook authentication required: HTTP webhook secret is not configured."
+                            .to_string(),
+                    ),
+                }),
+            )
+                .into_response();
+        };
+        let expected_secret = expected_secret.expose_secret();

-            match headers.get("x-ironclaw-signature") {
-                Some(raw_signature) => match raw_signature.to_str() {
-                    Ok(signature) => {
-                        if !verify_hmac_signature(expected_secret, &body, signature) {
-                            return (
-                                StatusCode::UNAUTHORIZED,
-                                Json(WebhookResponse {
-                                    message_id: Uuid::nil(),
-                                    status: "error".to_string(),
-                                    response: Some("Invalid webhook signature".to_string()),
-                                }),
-                            )
-                                .into_response();
-                        }
+        match headers.get("x-ironclaw-signature") {
+            Some(raw_signature) => match raw_signature.to_str() {
+                Ok(signature) => {
+                    if !verify_hmac_signature(expected_secret, &body, signature) {
+                        return (
+                            StatusCode::UNAUTHORIZED,
+                            Json(WebhookResponse {
+                                message_id: Uuid::nil(),
+                                status: "error".to_string(),
+                                response: Some("Invalid webhook signature".to_string()),
+                            }),
+                        )
+                            .into_response();
                    }
+                }
+                Err(_) => {
+                    return (
+                        StatusCode::UNAUTHORIZED,
+                        Json(WebhookResponse {
+                            message_id: Uuid::nil(),
+                            status: "error".to_string(),
+                            response: Some("Invalid signature header encoding".to_string()),
+                        }),
+                    )
+                        .into_response();
+                }
+            },
+            None => {
+                let req: WebhookRequest = match serde_json::from_slice(&body) {
+                    Ok(req) => req,
                    Err(_) => {
                        return (
                            StatusCode::UNAUTHORIZED,
                            Json(WebhookResponse {
                                message_id: Uuid::nil(),
                                status: "error".to_string(),
-                                response: Some("Invalid signature header encoding".to_string()),
+                                response: Some(
+                                    "Webhook authentication required. Provide X-IronClaw-Signature header \
+                                     (preferred) or 'secret' field in body (deprecated)."
+                                        .to_string(),
+                                ),
                            }),
                        )
                            .into_response();
                    }
-                },
-                None => {
-                    let req: WebhookRequest = match serde_json::from_slice(&body) {
-                        Ok(req) => req,
-                        Err(_) => {
-                            return (
-                                StatusCode::UNAUTHORIZED,
-                                Json(WebhookResponse {
-                                    message_id: Uuid::nil(),
-                                    status: "error".to_string(),
-                                    response: Some(
-                                        "Webhook authentication required. Provide X-IronClaw-Signature header \
-                                         (preferred) or 'secret' field in body (deprecated)."
-                                            .to_string(),
-                                    ),
-                                }),
-                            )
-                                .into_response();
-                        }
-                    };
+                };

-                    match &req.secret {
-                        Some(provided)
-                            if bool::from(
-                                provided.as_bytes().ct_eq(expected_secret.as_bytes()),
-                            ) =>
-                        {
-                            tracing::warn!(
-                                "Webhook authenticated via deprecated 'secret' field in request body. \
-                                 Migrate to X-IronClaw-Signature header (HMAC-SHA256). \
-                                 Body secret support will be removed in a future release."
-                            );
-                            fallback_req = Some(req);
-                        }
-                        Some(_) => {
-                            return (
-                                StatusCode::UNAUTHORIZED,
-                                Json(WebhookResponse {
-                                    message_id: Uuid::nil(),
-                                    status: "error".to_string(),
-                                    response: Some("Invalid webhook secret".to_string()),
-                                }),
-                            )
-                                .into_response();
-                        }
-                        None => {
-                            return (
-                                StatusCode::UNAUTHORIZED,
-                                Json(WebhookResponse {
-                                    message_id: Uuid::nil(),
-                                    status: "error".to_string(),
-                                    response: Some(
-                                        "Webhook authentication required. Provide X-IronClaw-Signature header \
-                                         (preferred) or 'secret' field in body (deprecated)."
-                                            .to_string(),
-                                    ),
-                                }),
-                            )
-                                .into_response();
-                        }
+                match &req.secret {
+                    Some(provided)
+                        if bool::from(provided.as_bytes().ct_eq(expected_secret.as_bytes())) =>
+                    {
+                        tracing::warn!(
+                            "Webhook authenticated via deprecated 'secret' field in request body. \
+                             Migrate to X-IronClaw-Signature header (HMAC-SHA256). \
+                             Body secret support will be removed in a future release."
+                        );
+                        fallback_req = Some(req);
+                    }
+                    Some(_) => {
+                        return (
+                            StatusCode::UNAUTHORIZED,
+                            Json(WebhookResponse {
+                                message_id: Uuid::nil(),
+                                status: "error".to_string(),
+                                response: Some("Invalid webhook secret".to_string()),
+                            }),
+                        )
+                            .into_response();
+                    }
+                    None => {
+                        return (
+                            StatusCode::UNAUTHORIZED,
+                            Json(WebhookResponse {
+                                message_id: Uuid::nil(),
+                                status: "error".to_string(),
+                                response: Some(
+                                    "Webhook authentication required. Provide X-IronClaw-Signature header \
+                                     (preferred) or 'secret' field in body (deprecated)."
+                                        .to_string(),
+                                ),
+                            }),
+                        )
+                            .into_response();
                    }
                }
            }
@@ -807,6 +817,67 @@ mod tests {
        assert_eq!(resp.status(), StatusCode::UNAUTHORIZED);
    }

+    /// Regression test for issue #869: RwLock read guard was held across
+    /// tx.send(msg).await in `process_message()`, blocking shutdown() from
+    /// acquiring the write lock when the channel buffer was full.
+    ///
+    /// This test exercises the actual production code path (`process_message`)
+    /// with a full channel buffer, then verifies shutdown() can still complete.
+    #[tokio::test]
+    async fn shutdown_completes_while_process_message_blocked() {
+        let channel = Arc::new(test_channel(Some("secret")));
+        let stream = channel.start().await.unwrap();
+
+        // Fill all 256 slots in the channel buffer
+        {
+            let tx = {
+                let guard = channel.state.tx.read().await;
+                guard.as_ref().unwrap().clone()
+            };
+            for i in 0..256 {
+                let msg = IncomingMessage::new("http", "user", format!("fill-{}", i));
+                tx.send(msg).await.unwrap();
+            }
+        }
+
+        // Signal so we know the spawned task has started and is about to
+        // call process_message (which will block on the full channel).
+        let started = Arc::new(tokio::sync::Notify::new());
+        let started_clone = started.clone();
+
+        // Spawn a task that calls the actual production code path.
+        // process_message() internally acquires the RwLock read guard and
+        // sends on the channel. With the fix, the guard is released before
+        // send().await; without the fix, shutdown() would deadlock.
+        let state = channel.state.clone();
+        let blocked_send = tokio::spawn(async move {
+            started_clone.notify_one();
+            let msg = IncomingMessage::new("http", "user", "blocked-257th");
+            let _ = process_message(state, msg, false).await;
+        });
+
+        // Wait for the spawned task to start, then give it time to reach
+        // the send().await and verify that it is still pending (i.e., blocked).
+        started.notified().await;
+        tokio::time::sleep(std::time::Duration::from_millis(50)).await;
+        assert!(
+            !blocked_send.is_finished(),
+            "process_message task should still be pending before shutdown()"
+        );
+
+        // shutdown() must complete even though process_message is blocked on
+        // send(). Before the fix, the read guard held across send().await
+        // would prevent shutdown() from acquiring the write lock.
+        let result =
+            tokio::time::timeout(std::time::Duration::from_secs(2), channel.shutdown()).await;
+        assert!(result.is_ok(), "shutdown() must not deadlock");
+        assert!(result.unwrap().is_ok());
+
+        // Drop the stream (receiver) so the blocked send task can complete
+        drop(stream);
+        let _ = blocked_send.await;
+    }
+
    #[tokio::test]
    async fn webhook_missing_all_auth_returns_unauthorized() {
        let channel = test_channel(Some("correct-secret"));
@@ -991,6 +1062,32 @@ mod tests {
        );
    }

+    #[tokio::test]
+    async fn webhook_rejects_requests_after_secret_is_cleared() {
+        let secret = "test-secret-123";
+        let channel = test_channel(Some(secret));
+        let _stream = channel.start().await.unwrap();
+        let app = channel.routes();
+
+        channel.update_secret(None).await;
+
+        let body = serde_json::json!({
+            "content": "hello"
+        });
+        let body_bytes = serde_json::to_vec(&body).unwrap();
+        let signature = compute_signature(secret, &body_bytes);
+        let req = Request::builder()
+            .method("POST")
+            .uri("/webhook")
+            .header("content-type", "application/json")
+            .header("x-ironclaw-signature", signature)
+            .body(Body::from(body_bytes))
+            .unwrap();
+
+        let resp = app.oneshot(req).await.unwrap();
+        assert_eq!(resp.status(), StatusCode::UNAUTHORIZED);
+    }
+
    #[tokio::test]
    async fn test_concurrent_requests_during_secret_update() {
        use std::sync::Arc as StdArc;
@@ -408,12 +408,120 @@ impl Channel for RelayChannel {
        Ok(())
    }

-    /// Status updates are not forwarded to messaging providers to avoid noise.
    async fn send_status(
        &self,
-        _status: StatusUpdate,
-        _metadata: &serde_json::Value,
+        status: StatusUpdate,
+        metadata: &serde_json::Value,
    ) -> Result<(), ChannelError> {
+        // Only handle ApprovalNeeded — all other variants are no-ops
+        let StatusUpdate::ApprovalNeeded {
+            request_id,
+            tool_name,
+            description,
+            parameters,
+        } = status
+        else {
+            return Ok(());
+        };
+
+        // Only send buttons in DMs (dispatcher gates upstream, but guard here too)
+        let event_type = metadata
+            .get("event_type")
+            .and_then(|v| v.as_str())
+            .unwrap_or("");
+        if event_type != "direct_message" {
+            tracing::warn!(
+                tool = %tool_name,
+                event_type,
+                "Approval requested in non-DM, skipping buttons"
+            );
+            return Ok(());
+        }
+
+        // Extract required metadata — error if missing
+        let channel_id = metadata
+            .get("channel_id")
+            .and_then(|v| v.as_str())
+            .ok_or_else(|| ChannelError::SendFailed {
+                name: self.name().to_string(),
+                reason: "Missing channel_id for approval buttons".into(),
+            })?;
+        let sender_id = metadata
+            .get("sender_id")
+            .and_then(|v| v.as_str())
+            .ok_or_else(|| ChannelError::SendFailed {
+                name: self.name().to_string(),
+                reason: "Missing sender_id for approval buttons".into(),
+            })?;
+        let thread_id = metadata.get("thread_id").and_then(|v| v.as_str());
+        let team_id = metadata
+            .get("team_id")
+            .and_then(|v| v.as_str())
+            .unwrap_or(&self.team_id);
+
+        // Button value payload (Slack limits button values to 2000 chars;
+        // safe with typical UUIDs but documented here as a constraint)
+        let value_payload = serde_json::json!({
+            "instance_id": self.instance_id,
+            "team_id": team_id,
+            "channel_id": channel_id,
+            "thread_ts": thread_id,
+            "request_id": request_id,
+            "sender_id": sender_id,
+        });
+        let value_str = value_payload.to_string();
+
+        // Parameters are already redacted via redact_params() in dispatcher.rs
+        let params_display =
+            serde_json::to_string_pretty(&parameters).unwrap_or_else(|_| parameters.to_string());
+
+        let blocks = serde_json::json!([
+            {
+                "type": "section",
+                "text": {
+                    "type": "mrkdwn",
+                    "text": format!(
+                        "*Tool approval required*\n`{tool_name}`: {description}\n```{params_display}```"
+                    )
+                }
+            },
+            {
+                "type": "actions",
+                "elements": [
+                    {
+                        "type": "button",
+                        "text": { "type": "plain_text", "text": "Approve" },
+                        "style": "primary",
+                        "action_id": "approve_tool",
+                        "value": value_str,
+                    },
+                    {
+                        "type": "button",
+                        "text": { "type": "plain_text", "text": "Deny" },
+                        "style": "danger",
+                        "action_id": "deny_tool",
+                        "value": value_str,
+                    }
+                ]
+            }
+        ]);
+
+        let mut body = serde_json::json!({
+            "channel": channel_id,
+            "text": format!("Tool approval required: {tool_name} - {description}"),
+            "blocks": blocks,
+        });
+        if let Some(tid) = thread_id {
+            body["thread_ts"] = serde_json::Value::String(tid.to_string());
+        }
+
+        self.proxy_send(team_id, "chat.postMessage", body)
+            .await
+            .map_err(|e| ChannelError::SendFailed {
+                name: self.name().to_string(),
+                reason: e.to_string(),
+            })?;
+
        Ok(())
    }

@@ -639,4 +747,118 @@ mod tests {
        // The reconnect loop now skips team validation when team_id is empty,
        // so the channel remains alive.
    }
+
+    #[tokio::test]
+    async fn test_send_status_non_approval_is_noop() {
+        let channel = RelayChannel::new(
+            test_client(),
+            "token".into(),
+            "T123".into(),
+            "inst1".into(),
+            "user1".into(),
+        );
+        let metadata = serde_json::json!({});
+        let result = channel
+            .send_status(
+                StatusUpdate::ToolStarted {
+                    name: "echo".into(),
+                },
+                &metadata,
+            )
+            .await;
+        assert!(result.is_ok());
+    }
+
+    #[tokio::test]
+    async fn test_send_status_approval_non_dm_skips() {
+        let channel = RelayChannel::new(
+            test_client(),
+            "token".into(),
+            "T123".into(),
+            "inst1".into(),
+            "user1".into(),
+        );
+        let metadata = serde_json::json!({
+            "event_type": "message",
+            "channel_id": "C456",
+            "sender_id": "U789",
+        });
+        let result = channel
+            .send_status(
+                StatusUpdate::ApprovalNeeded {
+                    request_id: "req1".into(),
+                    tool_name: "shell".into(),
+                    description: "run command".into(),
+                    parameters: serde_json::json!({}),
+                },
+                &metadata,
+            )
+            .await;
+        // Non-DM approval requests are silently skipped (no HTTP call)
+        assert!(result.is_ok());
+    }
+
+    #[tokio::test]
+    async fn test_send_status_approval_dm_missing_channel_id_errors() {
+        let channel = RelayChannel::new(
+            test_client(),
+            "token".into(),
+            "T123".into(),
+            "inst1".into(),
+            "user1".into(),
+        );
+        let metadata = serde_json::json!({
+            "event_type": "direct_message",
+            "sender_id": "U789",
+        });
+        let result = channel
+            .send_status(
+                StatusUpdate::ApprovalNeeded {
+                    request_id: "req1".into(),
+                    tool_name: "shell".into(),
+                    description: "run command".into(),
+                    parameters: serde_json::json!({}),
+                },
+                &metadata,
+            )
+            .await;
+        assert!(result.is_err());
+        let err = result.unwrap_err().to_string();
+        assert!(
+            err.contains("channel_id"),
+            "expected channel_id error, got: {err}"
+        );
+    }
+
+    #[tokio::test]
+    async fn test_send_status_approval_dm_missing_sender_id_errors() {
+        let channel = RelayChannel::new(
+            test_client(),
+            "token".into(),
+            "T123".into(),
+            "inst1".into(),
+            "user1".into(),
+        );
+        let metadata = serde_json::json!({
+            "event_type": "direct_message",
+            "channel_id": "C456",
+        });
+        let result = channel
+            .send_status(
+                StatusUpdate::ApprovalNeeded {
+                    request_id: "req1".into(),
+                    tool_name: "shell".into(),
+                    description: "run command".into(),
+                    parameters: serde_json::json!({}),
+                },
+                &metadata,
+            )
+            .await;
+        assert!(result.is_err());
+        let err = result.unwrap_err().to_string();
+        assert!(
+            err.contains("sender_id"),
+            "expected sender_id error, got: {err}"
+        );
+    }
 }
@@ -63,7 +63,11 @@ const ALLOWED_MIME_PREFIXES: &[&str] = &[
    "application/x-tar",
    "application/octet-stream",
 ];
-
+/// Truncate a string to at most `max_bytes` without splitting UTF-8 code points.
+fn truncate_utf8(s: &str, max_bytes: usize) -> &str {
+    let end = crate::util::floor_char_boundary(s, max_bytes);
+    &s[..end]
+}
 /// A message emitted by a WASM channel to be sent to the agent.
 #[derive(Debug, Clone)]
 pub struct EmittedMessage {
@@ -264,7 +268,7 @@ impl ChannelHostState {
                max = MAX_MESSAGE_CONTENT_SIZE,
                "Message content too large, truncating"
            );
-            let mut truncated = msg.content[..MAX_MESSAGE_CONTENT_SIZE].to_string();
+            let mut truncated = truncate_utf8(&msg.content, MAX_MESSAGE_CONTENT_SIZE).to_string();
            truncated.push_str("... (truncated)");
            let msg = EmittedMessage {
                content: truncated,
@@ -631,6 +635,7 @@ mod tests {
    use crate::channels::wasm::host::{
        Attachment, ChannelEmitRateLimiter, ChannelHostState, EmittedMessage,
        MAX_ATTACHMENT_TOTAL_SIZE, MAX_ATTACHMENTS_PER_MESSAGE, MAX_EMITS_PER_EXECUTION,
+        MAX_MESSAGE_CONTENT_SIZE,
    };

    #[test]
@@ -689,6 +694,25 @@ mod tests {
        assert_eq!(state.emits_dropped(), 1);
    }

+    #[test]
+    fn test_emit_message_truncates_utf8_safely() {
+        let caps = ChannelCapabilities::for_channel("test");
+        let mut state = ChannelHostState::new("test", caps);
+
+        let prefix = "a".repeat(MAX_MESSAGE_CONTENT_SIZE - 1);
+        let content = format!("{}🙂suffix", prefix);
+        let msg = EmittedMessage::new("user123", content);
+
+        state.emit_message(msg).unwrap();
+        let messages = state.take_emitted_messages();
+        assert_eq!(messages.len(), 1);
+
+        let emitted = &messages[0].content;
+        assert!(emitted.starts_with(&prefix));
+        assert!(emitted.ends_with("... (truncated)"));
+        assert!(!emitted.contains("🙂"));
+    }
+
    #[test]
    fn test_workspace_write_prefixing() {
        let caps = ChannelCapabilities::for_channel("slack");
@@ -1994,28 +1994,33 @@ impl WasmChannel {
            return Ok(());
        }

-        let tx_guard = self.message_tx.read().await;
-        let Some(tx) = tx_guard.as_ref() else {
-            tracing::error!(
-                channel = %self.name,
-                count = messages.len(),
-                "Messages emitted but no sender available - channel may not be started!"
-            );
-            return Ok(());
+        // Clone sender to avoid holding RwLock read guard across send().await in the loop
+        let tx = {
+            let tx_guard = self.message_tx.read().await;
+            let Some(tx) = tx_guard.as_ref() else {
+                tracing::error!(
+                    channel = %self.name,
+                    count = messages.len(),
+                    "Messages emitted but no sender available - channel may not be started!"
+                );
+                return Ok(());
+            };
+            tx.clone()
        };

-        let mut rate_limiter = self.rate_limiter.write().await;
-
        for emitted in messages {
-            // Check rate limit
-            if !rate_limiter.check_and_record() {
-                tracing::warn!(
-                    channel = %self.name,
-                    "Message emission rate limited"
-                );
-                return Err(WasmChannelError::EmitRateLimited {
-                    name: self.name.clone(),
-                });
+            // Check rate limit — acquire and release the write lock before send().await
+            {
+                let mut rate_limiter = self.rate_limiter.write().await;
+                if !rate_limiter.check_and_record() {
+                    tracing::warn!(
+                        channel = %self.name,
+                        "Message emission rate limited"
+                    );
+                    return Err(WasmChannelError::EmitRateLimited {
+                        name: self.name.clone(),
+                    });
+                }
            }

            // Convert to IncomingMessage
@@ -2057,7 +2062,7 @@ impl WasmChannel {
                self.update_broadcast_metadata(&emitted.metadata_json).await;
            }

-            // Send to stream
+            // Send to stream — no locks held across this await
            tracing::info!(
                channel = %self.name,
                user_id = %emitted.user_id,
@@ -2281,28 +2286,33 @@ impl WasmChannel {
            "Processing emitted messages from polling callback"
        );

-        let tx_guard = message_tx.read().await;
-        let Some(tx) = tx_guard.as_ref() else {
-            tracing::error!(
-                channel = %channel_name,
-                count = messages.len(),
-                "Messages emitted but no sender available - channel may not be started!"
-            );
-            return Ok(());
+        // Clone sender to avoid holding RwLock read guard across send().await in the loop
+        let tx = {
+            let tx_guard = message_tx.read().await;
+            let Some(tx) = tx_guard.as_ref() else {
+                tracing::error!(
+                    channel = %channel_name,
+                    count = messages.len(),
+                    "Messages emitted but no sender available - channel may not be started!"
+                );
+                return Ok(());
+            };
+            tx.clone()
        };

-        let mut limiter = rate_limiter.write().await;
-
        for emitted in messages {
-            // Check rate limit
-            if !limiter.check_and_record() {
-                tracing::warn!(
-                    channel = %channel_name,
-                    "Message emission rate limited"
-                );
-                return Err(WasmChannelError::EmitRateLimited {
-                    name: channel_name.to_string(),
-                });
+            // Check rate limit — acquire and release the write lock before send().await
+            {
+                let mut limiter = rate_limiter.write().await;
+                if !limiter.check_and_record() {
+                    tracing::warn!(
+                        channel = %channel_name,
+                        "Message emission rate limited"
+                    );
+                    return Err(WasmChannelError::EmitRateLimited {
+                        name: channel_name.to_string(),
+                    });
+                }
            }

            // Convert to IncomingMessage
@@ -2350,7 +2360,7 @@ impl WasmChannel {
                .await;
            }

-            // Send to stream
+            // Send to stream — no locks held across this await
            tracing::info!(
                channel = %channel_name,
                user_id = %emitted.user_id,
@@ -37,11 +37,17 @@ pub async fn chat_send_handler(
    let msg_id = msg.id;
    let thread_id = msg.thread_id.clone();

-    let tx_guard = state.msg_tx.read().await;
-    let tx = tx_guard.as_ref().ok_or((
-        StatusCode::SERVICE_UNAVAILABLE,
-        "Channel not started".to_string(),
-    ))?;
+    // Clone sender to avoid holding RwLock read guard across send().await
+    let tx = {
+        let tx_guard = state.msg_tx.read().await;
+        tx_guard
+            .as_ref()
+            .ok_or((
+                StatusCode::SERVICE_UNAVAILABLE,
+                "Channel not started".to_string(),
+            ))?
+            .clone()
+    };

    tx.send(msg).await.map_err(|_| {
        (
@@ -111,11 +117,17 @@ pub async fn chat_approval_handler(

    let msg_id = msg.id;

-    let tx_guard = state.msg_tx.read().await;
-    let tx = tx_guard.as_ref().ok_or((
-        StatusCode::SERVICE_UNAVAILABLE,
-        "Channel not started".to_string(),
-    ))?;
+    // Clone sender to avoid holding RwLock read guard across send().await
+    let tx = {
+        let tx_guard = state.msg_tx.read().await;
+        tx_guard
+            .as_ref()
+            .ok_or((
+                StatusCode::SERVICE_UNAVAILABLE,
+                "Channel not started".to_string(),
+            ))?
+            .clone()
+    };

    tx.send(msg).await.map_err(|_| {
        (
@@ -10,6 +10,7 @@ use axum::{
 use serde::Deserialize;
 use uuid::Uuid;

+use crate::agent::routine::{Trigger, next_cron_fire};
 use crate::channels::web::server::GatewayState;
 use crate::channels::web::types::*;
 use crate::error::RoutineError;
@@ -182,12 +183,21 @@ pub async fn routines_toggle_handler(
        .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?
        .ok_or((StatusCode::NOT_FOUND, "Routine not found".to_string()))?;

+    let was_enabled = routine.enabled;
    // If a specific value was provided, use it; otherwise toggle.
    routine.enabled = match body {
        Some(Json(req)) => req.enabled.unwrap_or(!routine.enabled),
        None => !routine.enabled,
    };

+    if routine.enabled
+        && !was_enabled
+        && let Trigger::Cron { schedule, timezone } = &routine.trigger
+    {
+        routine.next_fire_at = next_cron_fire(schedule, timezone.as_deref())
+            .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
+    }
+
    store
        .update_routine(&routine)
        .await
@@ -26,6 +26,7 @@ use tower_http::set_header::SetResponseHeaderLayer;
 use uuid::Uuid;

 use crate::agent::SessionManager;
+use crate::agent::routine::{Trigger, next_cron_fire};
 use crate::bootstrap::ironclaw_base_dir;
 use crate::channels::IncomingMessage;
 use crate::channels::relay::DEFAULT_RELAY_NAME;
@@ -572,6 +573,14 @@ async fn oauth_callback_handler(
            extension = %flow.extension_name,
            "OAuth flow expired"
        );
+        // Notify UI so auth card can show error instead of staying stuck
+        if let Some(ref sender) = flow.sse_sender {
+            let _ = sender.send(SseEvent::AuthCompleted {
+                extension_name: flow.extension_name.clone(),
+                success: false,
+                message: "OAuth flow expired. Please try again.".to_string(),
+            });
+        }
        return oauth_error_page(&flow.display_name);
    }

@@ -581,7 +590,12 @@ async fn oauth_callback_handler(
    let exchange_proxy_url = std::env::var("IRONCLAW_OAUTH_EXCHANGE_URL").ok();

    let result: Result<(), String> = async {
-        let token_response = if let Some(ref proxy_url) = exchange_proxy_url {
+        let token_response = if let (Some(proxy_url), None) = (&exchange_proxy_url, &flow.resource)
+        {
+            // Use the platform exchange proxy when configured and no resource
+            // parameter is needed. The proxy holds client_secret server-side so
+            // the container never sees it. MCP flows (resource.is_some()) bypass
+            // the proxy because it doesn't forward the RFC 8707 resource param.
            let gateway_token = flow.gateway_token.as_deref().unwrap_or_default();
            oauth_defaults::exchange_via_proxy(
                proxy_url,
@@ -594,7 +608,10 @@ async fn oauth_callback_handler(
            .await
            .map_err(|e| e.to_string())?
        } else {
-            oauth_defaults::exchange_oauth_code(
+            // Direct token exchange: uses exchange_oauth_code_with_resource so MCP
+            // flows can include the RFC 8707 `resource` parameter to scope the
+            // issued token to the specific MCP server.
+            oauth_defaults::exchange_oauth_code_with_resource(
                &flow.token_url,
                &flow.client_id,
                flow.client_secret.as_deref(),
@@ -602,6 +619,7 @@ async fn oauth_callback_handler(
                &flow.redirect_uri,
                flow.code_verifier.as_deref(),
                &flow.access_token_field,
+                flow.resource.as_deref(),
            )
            .await
            .map_err(|e| e.to_string())?
@@ -628,6 +646,19 @@ async fn oauth_callback_handler(
        .await
        .map_err(|e| e.to_string())?;

+        // For MCP OAuth flows (identified by resource field), persist the
+        // client_id so token refresh works without re-authentication.
+        // The CLI flow stores this in authorize_mcp_server(); the gateway
+        // callback must do the same.
+        if let Some(ref client_id_secret) = flow.client_id_secret_name {
+            let params = crate::secrets::CreateSecretParams::new(client_id_secret, &flow.client_id)
+                .with_provider(flow.provider.as_ref().cloned().unwrap_or_default());
+            flow.secrets
+                .create(&flow.user_id, params)
+                .await
+                .map_err(|e| e.to_string())?;
+        }
+
        Ok(())
    }
    .await;
@@ -659,12 +690,35 @@ async fn oauth_callback_handler(
        }
    }

+    // After successful OAuth, auto-activate the extension so it moves
+    // from "Installed (Authenticate)" → "Active" without a second click.
+    // OAuth success is independent of activation — tokens are already stored.
+    // Report auth as successful and attempt activation as a bonus step.
+    let final_message = if success {
+        match ext_mgr.activate(&flow.extension_name).await {
+            Ok(result) => result.message,
+            Err(e) => {
+                tracing::warn!(
+                    extension = %flow.extension_name,
+                    error = %e,
+                    "Auto-activation after OAuth failed"
+                );
+                format!(
+                    "{} authenticated successfully. Activation failed: {}. Try activating manually.",
+                    flow.display_name, e
+                )
+            }
+        }
+    } else {
+        message
+    };
+
    // Broadcast SSE event to notify the web UI
    if let Some(ref sender) = flow.sse_sender {
        let _ = sender.send(SseEvent::AuthCompleted {
            extension_name: flow.extension_name,
            success,
-            message,
+            message: final_message.clone(),
        });
    }

@@ -973,11 +1027,17 @@ async fn chat_send_handler(
        req.images.len()
    );

-    let tx_guard = state.msg_tx.read().await;
-    let tx = tx_guard.as_ref().ok_or((
-        StatusCode::SERVICE_UNAVAILABLE,
-        "Channel not started".to_string(),
-    ))?;
+    // Clone sender to avoid holding RwLock read guard across send().await
+    let tx = {
+        let tx_guard = state.msg_tx.read().await;
+        tx_guard
+            .as_ref()
+            .ok_or((
+                StatusCode::SERVICE_UNAVAILABLE,
+                "Channel not started".to_string(),
+            ))?
+            .clone()
+    };

    tracing::debug!("[chat_send_handler] Sending message through channel");
    tx.send(msg).await.map_err(|_| {
@@ -1043,11 +1103,17 @@ async fn chat_approval_handler(

    let msg_id = msg.id;

-    let tx_guard = state.msg_tx.read().await;
-    let tx = tx_guard.as_ref().ok_or((
-        StatusCode::SERVICE_UNAVAILABLE,
-        "Channel not started".to_string(),
-    ))?;
+    // Clone sender to avoid holding RwLock read guard across send().await
+    let tx = {
+        let tx_guard = state.msg_tx.read().await;
+        tx_guard
+            .as_ref()
+            .ok_or((
+                StatusCode::SERVICE_UNAVAILABLE,
+                "Channel not started".to_string(),
+            ))?
+            .clone()
+    };

    tx.send(msg).await.map_err(|_| {
        (
@@ -2359,12 +2425,21 @@ async fn routines_toggle_handler(
        .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?
        .ok_or((StatusCode::NOT_FOUND, "Routine not found".to_string()))?;

+    let was_enabled = routine.enabled;
    // If a specific value was provided, use it; otherwise toggle.
    routine.enabled = match body {
        Some(Json(req)) => req.enabled.unwrap_or(!routine.enabled),
        None => !routine.enabled,
    };

+    if routine.enabled
+        && !was_enabled
+        && let Trigger::Cron { schedule, timezone } = &routine.trigger
+    {
+        routine.next_fire_at = next_cron_fire(schedule, timezone.as_deref())
+            .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
+    }
+
    store
        .update_routine(&routine)
        .await
@@ -2639,6 +2714,7 @@ struct GatewayStatusResponse {
 #[cfg(test)]
 mod tests {
    use super::*;
+    use crate::cli::oauth_defaults;
    use crate::testing::credentials::TEST_GATEWAY_CRYPTO_KEY;

    #[test]
@@ -2756,6 +2832,11 @@ mod tests {
            .with_state(state)
    }

+    fn expired_flow_created_at() -> Option<std::time::Instant> {
+        std::time::Instant::now()
+            .checked_sub(oauth_defaults::OAUTH_FLOW_EXPIRY + std::time::Duration::from_secs(1))
+    }
+
    #[tokio::test]
    async fn test_csp_header_present_on_responses() {
        use std::net::SocketAddr;
@@ -2862,29 +2943,14 @@ mod tests {
        use tower::ServiceExt;

        // Build an ExtensionManager so the handler can look up flows
-        let secrets = Arc::new(crate::secrets::InMemorySecretsStore::new(Arc::new(
-            crate::secrets::SecretsCrypto::new(secrecy::SecretString::from(
-                TEST_GATEWAY_CRYPTO_KEY.to_string(),
-            ))
-            .expect("crypto"),
-        )));
-        let tool_registry = Arc::new(ToolRegistry::new());
-        let mcp_sm = Arc::new(crate::tools::mcp::session::McpSessionManager::new());
-
-        let ext_mgr = Arc::new(ExtensionManager::new(
-            mcp_sm,
-            Arc::new(crate::tools::mcp::process::McpProcessManager::new()),
-            secrets,
-            tool_registry,
-            None,
-            None,
-            std::path::PathBuf::from("/tmp/wasm_tools"),
-            std::path::PathBuf::from("/tmp/wasm_channels"),
-            None,
-            "test".to_string(),
-            None,
-            vec![],
-        ));
+        let secrets: Arc<dyn crate::secrets::SecretsStore + Send + Sync> =
+            Arc::new(crate::secrets::InMemorySecretsStore::new(Arc::new(
+                crate::secrets::SecretsCrypto::new(secrecy::SecretString::from(
+                    TEST_GATEWAY_CRYPTO_KEY.to_string(),
+                ))
+                .expect("crypto"),
+            )));
+        let (ext_mgr, _wasm_tools_dir, _wasm_channels_dir) = test_ext_mgr(secrets);

        let state = test_gateway_state(Some(ext_mgr));
        let app = test_oauth_router(state);
@@ -2918,25 +2984,13 @@ mod tests {
                ))
                .expect("crypto"),
            )));
-        let tool_registry = Arc::new(ToolRegistry::new());
-        let mcp_sm = Arc::new(crate::tools::mcp::session::McpSessionManager::new());
+        let (ext_mgr, _wasm_tools_dir, _wasm_channels_dir) = test_ext_mgr(secrets.clone());
+        let Some(created_at) = expired_flow_created_at() else {
+            eprintln!("Skipping expired OAuth flow test: monotonic uptime below expiry window");
+            return;
+        };

-        let ext_mgr = Arc::new(ExtensionManager::new(
-            mcp_sm,
-            Arc::new(crate::tools::mcp::process::McpProcessManager::new()),
-            secrets.clone(),
-            tool_registry,
-            None,
-            None,
-            std::path::PathBuf::from("/tmp/wasm_tools"),
-            std::path::PathBuf::from("/tmp/wasm_channels"),
-            None,
-            "test".to_string(),
-            None,
-            vec![],
-        ));
-
-        // Insert an expired flow (created 10 minutes ago)
+        // Insert an expired flow.
        let flow = crate::cli::oauth_defaults::PendingOAuthFlow {
            extension_name: "test_tool".to_string(),
            display_name: "Test Tool".to_string(),
@@ -2954,9 +3008,9 @@ mod tests {
            secrets,
            sse_sender: None,
            gateway_token: None,
-            created_at: std::time::Instant::now()
-                .checked_sub(std::time::Duration::from_secs(600))
-                .expect("System uptime is too low to run expired flow test"),
+            resource: None,
+            client_id_secret_name: None,
+            created_at,
        };

        ext_mgr
@@ -2986,6 +3040,80 @@ mod tests {
        assert!(html.contains("Authorization Failed"));
    }

+    #[tokio::test]
+    async fn test_oauth_callback_expired_flow_broadcasts_auth_completed_failure() {
+        use axum::body::Body;
+        use tower::ServiceExt;
+
+        let secrets: Arc<dyn crate::secrets::SecretsStore + Send + Sync> =
+            Arc::new(crate::secrets::InMemorySecretsStore::new(Arc::new(
+                crate::secrets::SecretsCrypto::new(secrecy::SecretString::from(
+                    TEST_GATEWAY_CRYPTO_KEY.to_string(),
+                ))
+                .expect("crypto"),
+            )));
+        let (ext_mgr, _wasm_tools_dir, _wasm_channels_dir) = test_ext_mgr(secrets.clone());
+
+        let (sender, mut receiver) = tokio::sync::broadcast::channel(4);
+        let Some(created_at) = expired_flow_created_at() else {
+            eprintln!("Skipping expired OAuth flow SSE test: monotonic uptime below expiry window");
+            return;
+        };
+        let flow = crate::cli::oauth_defaults::PendingOAuthFlow {
+            extension_name: "test_tool".to_string(),
+            display_name: "Test Tool".to_string(),
+            token_url: "https://example.com/token".to_string(),
+            client_id: "client123".to_string(),
+            client_secret: None,
+            redirect_uri: "https://example.com/oauth/callback".to_string(),
+            code_verifier: None,
+            access_token_field: "access_token".to_string(),
+            secret_name: "test_token".to_string(),
+            provider: None,
+            validation_endpoint: None,
+            scopes: vec![],
+            user_id: "test".to_string(),
+            secrets,
+            sse_sender: Some(sender),
+            gateway_token: None,
+            resource: None,
+            client_id_secret_name: None,
+            created_at,
+        };
+
+        ext_mgr
+            .pending_oauth_flows()
+            .write()
+            .await
+            .insert("expired_state".to_string(), flow);
+
+        let state = test_gateway_state(Some(ext_mgr));
+        let app = test_oauth_router(state);
+
+        let req = axum::http::Request::builder()
+            .uri("/oauth/callback?code=test_code&state=expired_state")
+            .body(Body::empty())
+            .expect("request");
+
+        let resp = ServiceExt::<axum::http::Request<Body>>::oneshot(app, req)
+            .await
+            .expect("response");
+        assert_eq!(resp.status(), StatusCode::OK);
+
+        match receiver.recv().await.expect("auth_completed event") {
+            crate::channels::web::types::SseEvent::AuthCompleted {
+                extension_name,
+                success,
+                message,
+            } => {
+                assert_eq!(extension_name, "test_tool");
+                assert!(!success, "expired OAuth flow should broadcast failure");
+                assert_eq!(message, "OAuth flow expired. Please try again.");
+            }
+            event => panic!("expected AuthCompleted event, got {event:?}"),
+        }
+    }
+
    #[tokio::test]
    async fn test_oauth_callback_no_extension_manager() {
        use axum::body::Body;
@@ -3024,28 +3152,16 @@ mod tests {
                ))
                .expect("crypto"),
            )));
-        let tool_registry = Arc::new(ToolRegistry::new());
-        let mcp_sm = Arc::new(crate::tools::mcp::session::McpSessionManager::new());
-
-        let ext_mgr = Arc::new(ExtensionManager::new(
-            mcp_sm,
-            Arc::new(crate::tools::mcp::process::McpProcessManager::new()),
-            secrets.clone(),
-            tool_registry,
-            None,
-            None,
-            std::path::PathBuf::from("/tmp/wasm_tools"),
-            std::path::PathBuf::from("/tmp/wasm_channels"),
-            None,
-            "test".to_string(),
-            None,
-            vec![],
-        ));
+        let (ext_mgr, _wasm_tools_dir, _wasm_channels_dir) = test_ext_mgr(secrets.clone());

        // Insert a flow keyed by raw nonce "test_nonce" (without instance prefix).
        // Use an expired flow so the handler exits before attempting a real HTTP
        // token exchange — we only need to verify that the instance prefix was
        // stripped and the flow was found by the raw nonce.
+        let Some(created_at) = expired_flow_created_at() else {
+            eprintln!("Skipping OAuth state-prefix test: monotonic uptime below expiry window");
+            return;
+        };
        let flow = crate::cli::oauth_defaults::PendingOAuthFlow {
            extension_name: "test_tool".to_string(),
            display_name: "Test Tool".to_string(),
@@ -3063,10 +3179,10 @@ mod tests {
            secrets,
            sse_sender: None,
            gateway_token: None,
+            resource: None,
+            client_id_secret_name: None,
            // Expired — handler will reject after lookup (no network I/O)
-            created_at: std::time::Instant::now()
-                .checked_sub(std::time::Duration::from_secs(600))
-                .expect("System uptime is too low to run expired flow test"),
+            created_at,
        };

        ext_mgr
@@ -3137,24 +3253,27 @@ mod tests {

    fn test_ext_mgr(
        secrets: Arc<dyn crate::secrets::SecretsStore + Send + Sync>,
-    ) -> Arc<ExtensionManager> {
+    ) -> (Arc<ExtensionManager>, tempfile::TempDir, tempfile::TempDir) {
        let tool_registry = Arc::new(ToolRegistry::new());
        let mcp_sm = Arc::new(crate::tools::mcp::session::McpSessionManager::new());
        let mcp_pm = Arc::new(crate::tools::mcp::process::McpProcessManager::new());
-        Arc::new(ExtensionManager::new(
+        let wasm_tools_dir = tempfile::tempdir().expect("temp wasm tools dir");
+        let wasm_channels_dir = tempfile::tempdir().expect("temp wasm channels dir");
+        let ext_mgr = Arc::new(ExtensionManager::new(
            mcp_sm,
            mcp_pm,
            secrets,
            tool_registry,
            None,
            None,
-            std::path::PathBuf::from("/tmp/wasm_tools"),
-            std::path::PathBuf::from("/tmp/wasm_channels"),
+            wasm_tools_dir.path().to_path_buf(),
+            wasm_channels_dir.path().to_path_buf(),
            None,
            "test".to_string(),
            None,
            vec![],
-        ))
+        ));
+        (ext_mgr, wasm_tools_dir, wasm_channels_dir)
    }

    #[tokio::test]
@@ -3163,7 +3282,7 @@ mod tests {
        use tower::ServiceExt;

        let secrets = test_secrets_store();
-        let ext_mgr = test_ext_mgr(secrets);
+        let (ext_mgr, _wasm_tools_dir, _wasm_channels_dir) = test_ext_mgr(secrets);
        let state = test_gateway_state(Some(ext_mgr));
        let app = test_relay_oauth_router(state);

@@ -3207,7 +3326,7 @@ mod tests {
            .await
            .expect("store nonce");

-        let ext_mgr = test_ext_mgr(secrets);
+        let (ext_mgr, _wasm_tools_dir, _wasm_channels_dir) = test_ext_mgr(secrets);
        let state = test_gateway_state(Some(ext_mgr));
        let app = test_relay_oauth_router(state);

@@ -3252,7 +3371,7 @@ mod tests {
            .await
            .expect("store nonce");

-        let ext_mgr = test_ext_mgr(secrets.clone());
+        let (ext_mgr, _wasm_tools_dir, _wasm_channels_dir) = test_ext_mgr(secrets.clone());
        let state = test_gateway_state(Some(ext_mgr));
        let app = test_relay_oauth_router(state);

@@ -342,31 +342,27 @@ function connectSSE() {

  eventSource.addEventListener('approval_needed', (e) => {
    const data = JSON.parse(e.data);
-    if (!isCurrentThread(data.thread_id)) return;
-    showApproval(data);
+    const hasThread = !!data.thread_id;
+    const forCurrentThread = !hasThread || isCurrentThread(data.thread_id);
+
+    if (forCurrentThread) {
+      showApproval(data);
+    } else {
+      // Keep thread list fresh when approval is requested in a background thread.
+      unreadThreads.set(data.thread_id, (unreadThreads.get(data.thread_id) || 0) + 1);
+      debouncedLoadThreads();
+    }
+
+    // Extension setup flows can surface approvals while user is on Extensions tab.
+    if (currentTab === 'extensions') loadExtensions();
  });

  eventSource.addEventListener('auth_required', (e) => {
-    const data = JSON.parse(e.data);
-    if (data.auth_url) {
-      // OAuth flow: show the auth card with an OAuth button + optional token paste field.
-      showAuthCard(data);
-    } else {
-      // Setup flow: fetch the extension's credential schema and show the multi-field
-      // configure modal (the same UI used by the Extensions tab "Setup" button).
-      showConfigureModal(data.extension_name);
-    }
+    handleAuthRequired(JSON.parse(e.data));
  });

  eventSource.addEventListener('auth_completed', (e) => {
-    const data = JSON.parse(e.data);
-    // Dismiss whichever UI path was active: auth card (OAuth) or configure modal (setup).
-    removeAuthCard(data.extension_name);
-    closeConfigureModal();
-    showToast(data.message, data.success ? 'success' : 'error');
-    // Refresh extensions list so status indicators update
-    if (currentTab === 'extensions') loadExtensions();
-    enableChatInput();
+    handleAuthCompleted(JSON.parse(e.data));
  });

  eventSource.addEventListener('extension_status', (e) => {
@@ -438,7 +434,8 @@ function sendMessage() {
  const content = input.value.trim();
  if (!content && stagedImages.length === 0) return;

-  addMessage('user', content || '(images attached)');
+  const imagesToDisplay = stagedImages.slice(); // snapshot before clearing
+  addMessage('user', content, imagesToDisplay);
  input.value = '';
  autoResizeTextarea(input);
  input.focus();
@@ -544,6 +541,24 @@ document.getElementById('chat-input').addEventListener('paste', (e) => {
  }
 });

+document.getElementById('chat-input').addEventListener('dragover', (e) => {
+  const hasImage = Array.from(e.dataTransfer.items || []).some(
+    item => item.kind === 'file' && item.type.startsWith('image/')
+  );
+  if (hasImage) {
+    e.preventDefault();
+    e.dataTransfer.dropEffect = 'copy';
+  }
+});
+
+document.getElementById('chat-input').addEventListener('drop', (e) => {
+  const files = Array.from(e.dataTransfer.files || []).filter(f => f.type.startsWith('image/'));
+  if (files.length > 0) {
+    e.preventDefault();
+    handleImageFiles(files);
+  }
+});
+
 function addGeneratedImage(dataUrl, path) {
  const container = document.getElementById('chat-messages');
  const card = document.createElement('div');
@@ -670,7 +685,7 @@ function renderMarkdown(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>');
+    html = html.replace(/<pre>/g, '<pre class="code-block-wrapper"><button class="copy-btn" data-action="copy-code">Copy</button>');
    return html;
  }
  return escapeHtml(text);
@@ -702,12 +717,44 @@ function copyCodeBlock(btn) {
  });
 }

-function addMessage(role, content) {
+function copyMessage(btn) {
+  const message = btn.closest('.message');
+  if (!message) return;
+  const text = message.getAttribute('data-copy-text')
+    || message.getAttribute('data-raw')
+    || message.textContent
+    || '';
+  navigator.clipboard.writeText(text).then(() => {
+    btn.textContent = 'Copied';
+    setTimeout(() => { btn.textContent = 'Copy'; }, 1200);
+  }).catch(() => {
+    btn.textContent = 'Failed';
+    setTimeout(() => { btn.textContent = 'Copy'; }, 1200);
+  });
+}
+
+function addMessage(role, content, images) {
  const container = document.getElementById('chat-messages');
  const div = document.createElement('div');
  div.className = 'message ' + role;
  if (role === 'user') {
-    div.textContent = content;
+    if (images && images.length > 0) {
+      const imgStrip = document.createElement('div');
+      imgStrip.className = 'message-images';
+      images.forEach(img => {
+        const thumb = document.createElement('img');
+        thumb.className = 'message-image-thumb';
+        thumb.src = img.dataUrl;
+        thumb.alt = 'Attached image';
+        imgStrip.appendChild(thumb);
+      });
+      div.appendChild(imgStrip);
+    }
+    if (content) {
+      const textDiv = document.createElement('div');
+      textDiv.textContent = content;
+      div.appendChild(textDiv);
+    }
  } else {
    div.setAttribute('data-raw', content);
    div.innerHTML = renderMarkdown(content);
@@ -723,7 +770,11 @@ function appendToLastAssistant(chunk) {
    const last = messages[messages.length - 1];
    const raw = (last.getAttribute('data-raw') || '') + chunk;
    last.setAttribute('data-raw', raw);
-    last.innerHTML = renderMarkdown(raw);
+    last.setAttribute('data-copy-text', raw);
+    const content = last.querySelector('.message-content');
+    if (content) {
+      content.innerHTML = renderMarkdown(raw);
+    }
    container.scrollTop = container.scrollHeight;
  } else {
    addMessage('assistant', chunk);
@@ -977,7 +1028,26 @@ function finalizeActivityGroup() {
  _activeToolCards = {};
 }

+function humanizeToolName(rawName) {
+  if (!rawName) return '';
+  return String(rawName)
+    .replace(/[_-]+/g, ' ')
+    .replace(/([a-z0-9])([A-Z])/g, '$1 $2')
+    .replace(/^tool([a-zA-Z])/, 'tool $1')
+    .replace(/\s+/g, ' ')
+    .trim();
+}
+
+function shouldShowChannelConnectedMessage(extensionName, success) {
+  if (!success || !extensionName) return false;
+  return String(extensionName).toLowerCase().includes('telegram');
+}
+
 function showApproval(data) {
+  // Avoid duplicate cards on reconnect/history refresh.
+  const existing = document.querySelector('.approval-card[data-request-id="' + CSS.escape(data.request_id) + '"]');
+  if (existing) return;
+
  const container = document.getElementById('chat-messages');
  const card = document.createElement('div');
  card.className = 'approval-card';
@@ -990,7 +1060,7 @@ function showApproval(data) {

  const toolName = document.createElement('div');
  toolName.className = 'approval-tool-name';
-  toolName.textContent = data.tool_name;
+  toolName.textContent = humanizeToolName(data.tool_name);
  card.appendChild(toolName);

  if (data.description) {
@@ -1093,13 +1163,71 @@ function showJobCard(data) {

 // --- Auth card ---

+function handleAuthRequired(data) {
+  if (data.auth_url) {
+    // OAuth flow: show the global auth prompt with an OAuth button + optional token paste field.
+    showAuthCard(data);
+  } else {
+    // Setup flow: fetch the extension's credential schema and show the multi-field
+    // configure modal (the same UI used by the Extensions tab "Setup" button).
+    showConfigureModal(data.extension_name);
+  }
+}
+
+function handleAuthCompleted(data) {
+  // Dismiss only the matching extension's UI so unrelated setup work is not interrupted.
+  removeAuthCard(data.extension_name);
+  closeConfigureModal(data.extension_name);
+  showToast(data.message, data.success ? 'success' : 'error');
+  if (shouldShowChannelConnectedMessage(data.extension_name, data.success)) {
+    addMessage('system', 'Telegram is now connected. You can message me there and I can send you notifications.');
+  }
+  if (currentTab === 'extensions') loadExtensions();
+  enableChatInput();
+}
+
+function queryByDataAttribute(selector, attributeName, attributeValue) {
+  if (typeof attributeValue !== 'string') return document.querySelector(selector);
+
+  if (window.CSS && typeof window.CSS.escape === 'function') {
+    return document.querySelector(
+      selector + '[' + attributeName + '="' + window.CSS.escape(attributeValue) + '"]'
+    );
+  }
+
+  const candidates = document.querySelectorAll(selector);
+  for (const candidate of candidates) {
+    if (candidate.getAttribute(attributeName) === attributeValue) return candidate;
+  }
+  return null;
+}
+
+function getAuthOverlay(extensionName) {
+  return queryByDataAttribute('.auth-overlay', 'data-extension-name', extensionName);
+}
+
+function getAuthCard(extensionName) {
+  return queryByDataAttribute('.auth-card', 'data-extension-name', extensionName);
+}
+
+function getConfigureOverlay(extensionName) {
+  return queryByDataAttribute('.configure-overlay', 'data-extension-name', extensionName);
+}
+
 function showAuthCard(data) {
-  // Remove any existing card for this extension first
-  removeAuthCard(data.extension_name);
+  // Keep a single global auth prompt so the experience is consistent across tabs.
+  const existing = getAuthOverlay();
+  if (existing) existing.remove();
+
+  const overlay = document.createElement('div');
+  overlay.className = 'auth-overlay';
+  overlay.setAttribute('data-extension-name', data.extension_name);
+  overlay.addEventListener('click', (e) => {
+    if (e.target === overlay) cancelAuth(data.extension_name);
+  });

-  const container = document.getElementById('chat-messages');
  const card = document.createElement('div');
-  card.className = 'auth-card';
+  card.className = 'auth-card auth-modal';
  card.setAttribute('data-extension-name', data.extension_name);

  const header = document.createElement('div');
@@ -1178,21 +1306,30 @@ function showAuthCard(data) {
  actions.appendChild(cancelBtn);
  card.appendChild(actions);

-  container.appendChild(card);
-  container.scrollTop = container.scrollHeight;
+  overlay.appendChild(card);
+  document.body.appendChild(overlay);
  tokenInput.focus();
 }

 function removeAuthCard(extensionName) {
-  const card = document.querySelector('.auth-card[data-extension-name="' + extensionName + '"]');
-  if (card) card.remove();
+  const overlay = getAuthOverlay(extensionName);
+  if (overlay) {
+    overlay.remove();
+    return;
+  }
+  const card = getAuthCard(extensionName);
+  if (card) {
+    const parentOverlay = card.closest('.auth-overlay');
+    if (parentOverlay) parentOverlay.remove();
+    else card.remove();
+  }
 }

 function submitAuthToken(extensionName, tokenValue) {
  if (!tokenValue || !tokenValue.trim()) return;

  // Disable submit button while in flight
-  const card = document.querySelector('.auth-card[data-extension-name="' + extensionName + '"]');
+  const card = getAuthCard(extensionName);
  if (card) {
    const btns = card.querySelectorAll('button');
    btns.forEach((b) => { b.disabled = true; });
@@ -1203,8 +1340,10 @@ function submitAuthToken(extensionName, tokenValue) {
    body: { extension_name: extensionName, token: tokenValue.trim() },
  }).then((result) => {
    if (result.success) {
+      // Close immediately for responsiveness; the authoritative success UX
+      // (toast + extensions refresh) still comes from auth_completed SSE.
      removeAuthCard(extensionName);
-      addMessage('system', result.message);
+      enableChatInput();
    } else {
      showAuthCardError(extensionName, result.message);
    }
@@ -1223,7 +1362,7 @@ function cancelAuth(extensionName) {
 }

 function showAuthCardError(extensionName, message) {
-  const card = document.querySelector('.auth-card[data-extension-name="' + extensionName + '"]');
+  const card = getAuthCard(extensionName);
  if (!card) return;
  // Re-enable buttons
  const btns = card.querySelectorAll('button');
@@ -1310,12 +1449,31 @@ function loadHistory(before) {
 function createMessageElement(role, content) {
  const div = document.createElement('div');
  div.className = 'message ' + role;
-  if (role === 'user') {
-    div.textContent = content;
+
+  if (role === 'assistant' || role === 'user') {
+    div.classList.add('has-copy');
+    div.setAttribute('data-copy-text', content);
+    const copyBtn = document.createElement('button');
+    copyBtn.className = 'message-copy-btn';
+    copyBtn.type = 'button';
+    copyBtn.setAttribute('aria-label', 'Copy message');
+    copyBtn.textContent = 'Copy';
+    copyBtn.addEventListener('click', (e) => {
+      e.stopPropagation();
+      copyMessage(copyBtn);
+    });
+    div.appendChild(copyBtn);
+  }
+
+  const body = document.createElement('div');
+  body.className = 'message-content';
+  if (role === 'user' || role === 'system') {
+    body.textContent = content;
  } else {
    div.setAttribute('data-raw', content);
-    div.innerHTML = renderMarkdown(content);
+    body.innerHTML = renderMarkdown(content);
  }
+  div.appendChild(body);
  return div;
 }

@@ -1819,13 +1977,11 @@ function saveMemoryEdit() {

 function buildBreadcrumb(path) {
  const parts = path.split('/');
-  let html = '<a onclick="loadMemoryTree()">workspace</a>';
+  let html = '<a data-action="breadcrumb-root" href="#">workspace</a>';
  let current = '';
  for (const part of parts) {
    current += (current ? '/' : '') + part;
-    // Store the path in data-path (HTML-escaped) and read it back via this.dataset.path
-    // to avoid single-quote injection in inline JS string literals.
-    html += ' / <a onclick="readMemoryFile(this.dataset.path)" data-path="' + escapeHtml(current) + '">' + escapeHtml(part) + '</a>';
+    html += ' / <a data-action="breadcrumb-file" data-path="' + escapeHtml(current) + '" href="#">' + escapeHtml(part) + '</a>';
  }
  return html;
 }
@@ -2136,6 +2292,10 @@ function renderAvailableExtensionCard(entry) {
        showToast(I18n.t('extensions.installedSuccess', {name: entry.display_name}), 'success');
        // OAuth popup if auth started during install (builtin creds)
        if (res.auth_url) {
+          showAuthCard({
+            extension_name: entry.name,
+            auth_url: res.auth_url,
+          });
          showToast('Opening authentication for ' + entry.display_name, 'info');
          openOAuthUrl(res.auth_url);
        }
@@ -2401,6 +2561,10 @@ function activateExtension(name) {
      if (res.success) {
        // Even on success, the tool may need OAuth (e.g., WASM loaded but no token yet)
        if (res.auth_url) {
+          showAuthCard({
+            extension_name: name,
+            auth_url: res.auth_url,
+          });
          showToast('Opening authentication for ' + name, 'info');
          openOAuthUrl(res.auth_url);
        }
@@ -2409,6 +2573,10 @@ function activateExtension(name) {
      }

      if (res.auth_url) {
+        showAuthCard({
+          extension_name: name,
+          auth_url: res.auth_url,
+        });
        showToast('Opening authentication for ' + name, 'info');
        openOAuthUrl(res.auth_url);
      } else if (res.awaiting_token) {
@@ -2451,6 +2619,7 @@ function renderConfigureModal(name, secrets) {
  closeConfigureModal();
  const overlay = document.createElement('div');
  overlay.className = 'configure-overlay';
+  overlay.setAttribute('data-extension-name', name);
  overlay.addEventListener('click', (e) => {
    if (e.target === overlay) closeConfigureModal();
  });
@@ -2544,7 +2713,8 @@ function submitConfigureModal(name, fields) {
  }

  // Disable buttons to prevent double-submit
-  var btns = document.querySelectorAll('.configure-actions button');
+  const overlay = getConfigureOverlay(name) || document.querySelector('.configure-overlay');
+  var btns = overlay ? overlay.querySelectorAll('.configure-actions button') : [];
  btns.forEach(function(b) { b.disabled = true; });

  apiFetch('/api/extensions/' + encodeURIComponent(name) + '/setup', {
@@ -2555,8 +2725,10 @@ function submitConfigureModal(name, fields) {
      if (res.success) {
        closeConfigureModal();
        if (res.auth_url) {
-          // OAuth flow started — open consent popup. The auth_completed SSE will
-          // not arrive immediately (it fires after OAuth callback), so show a toast now.
+          showAuthCard({
+            extension_name: name,
+            auth_url: res.auth_url,
+          });
          showToast('Opening OAuth authorization for ' + name, 'info');
          openOAuthUrl(res.auth_url);
          loadExtensions();
@@ -2575,8 +2747,9 @@ function submitConfigureModal(name, fields) {
    });
 }

-function closeConfigureModal() {
-  const existing = document.querySelector('.configure-overlay');
+function closeConfigureModal(extensionName) {
+  if (typeof extensionName !== 'string') extensionName = null;
+  const existing = getConfigureOverlay(extensionName);
  if (existing) existing.remove();
 }

@@ -2795,11 +2968,11 @@ function renderJobsList(jobs) {

    let actionBtns = '';
    if (job.state === 'pending' || job.state === 'in_progress') {
-      actionBtns = '<button class="btn-cancel" onclick="event.stopPropagation(); cancelJob(\'' + job.id + '\')">Cancel</button>';
+      actionBtns = '<button class="btn-cancel" data-action="cancel-job" data-id="' + escapeHtml(job.id) + '">Cancel</button>';
    }
    // Retry is only shown in the detail view where can_restart is available.

-    return '<tr class="job-row" onclick="openJobDetail(\'' + job.id + '\')">'
+    return '<tr class="job-row" data-action="open-job" data-id="' + escapeHtml(job.id) + '">'
      + '<td title="' + escapeHtml(job.id) + '">' + shortId + '</td>'
      + '<td>' + escapeHtml(job.title) + '</td>'
      + '<td><span class="badge ' + stateClass + '">' + escapeHtml(job.state) + '</span></td>'
@@ -2862,12 +3035,12 @@ function renderJobDetail(job) {
  const header = document.createElement('div');
  header.className = 'job-detail-header';

-  let headerHtml = '<button class="btn-back" onclick="closeJobDetail()">&larr; Back</button>'
+  let headerHtml = '<button class="btn-back" data-action="close-job-detail">&larr; Back</button>'
    + '<h2>' + escapeHtml(job.title) + '</h2>'
    + '<span class="badge ' + stateClass + '">' + escapeHtml(job.state) + '</span>';

  if ((job.state === 'failed' || job.state === 'interrupted') && job.can_restart === true) {
-    headerHtml += '<button class="btn-restart" onclick="restartJob(\'' + job.id + '\')">Retry</button>';
+    headerHtml += '<button class="btn-restart" data-action="restart-job" data-id="' + escapeHtml(job.id) + '">Retry</button>';
  }
  if (job.browse_url) {
    headerHtml += '<a class="btn-browse" href="' + escapeHtml(job.browse_url) + '" target="_blank">Browse Files</a>';
@@ -3324,7 +3497,7 @@ function renderRoutinesList(routines) {
    const toggleLabel = r.enabled ? 'Disable' : 'Enable';
    const toggleClass = r.enabled ? 'btn-cancel' : 'btn-restart';

-    return '<tr class="routine-row" onclick="openRoutineDetail(\'' + r.id + '\')">'
+    return '<tr class="routine-row" data-action="open-routine" data-id="' + escapeHtml(r.id) + '">'
      + '<td>' + escapeHtml(r.name) + '</td>'
      + '<td>' + escapeHtml(r.trigger_summary) + '</td>'
      + '<td>' + escapeHtml(r.action_type) + '</td>'
@@ -3333,9 +3506,9 @@ function renderRoutinesList(routines) {
      + '<td>' + r.run_count + '</td>'
      + '<td><span class="badge ' + statusClass + '">' + escapeHtml(r.status) + '</span></td>'
      + '<td>'
-      + '<button class="' + toggleClass + '" onclick="event.stopPropagation(); toggleRoutine(\'' + r.id + '\')">' + toggleLabel + '</button> '
-      + '<button class="btn-restart" onclick="event.stopPropagation(); triggerRoutine(\'' + r.id + '\')">Run</button> '
-      + '<button class="btn-cancel" onclick="event.stopPropagation(); deleteRoutine(\'' + r.id + '\', \'' + escapeHtml(r.name) + '\')">Delete</button>'
+      + '<button class="' + toggleClass + '" data-action="toggle-routine" data-id="' + escapeHtml(r.id) + '">' + toggleLabel + '</button> '
+      + '<button class="btn-restart" data-action="trigger-routine" data-id="' + escapeHtml(r.id) + '">Run</button> '
+      + '<button class="btn-cancel" data-action="delete-routine" data-id="' + escapeHtml(r.id) + '" data-name="' + escapeHtml(r.name) + '">Delete</button>'
      + '</td>'
      + '</tr>';
  }).join('');
@@ -3371,7 +3544,7 @@ function renderRoutineDetail(routine) {
    : 'active';

  let html = '<div class="job-detail-header">'
-    + '<button class="btn-back" onclick="closeRoutineDetail()">&larr; Back</button>'
+    + '<button class="btn-back" data-action="close-routine-detail">&larr; Back</button>'
    + '<h2>' + escapeHtml(routine.name) + '</h2>'
    + '<span class="badge ' + statusClass + '">' + escapeHtml(statusLabel) + '</span>'
    + '</div>';
@@ -3418,7 +3591,7 @@ function renderRoutineDetail(routine) {
        + '<td>' + formatDate(run.completed_at) + '</td>'
        + '<td><span class="badge ' + runStatusClass + '">' + escapeHtml(run.status) + '</span></td>'
        + '<td>' + escapeHtml(run.result_summary || '-')
-          + (run.job_id ? ' <a href="#" onclick="event.preventDefault(); switchTab(\'jobs\'); openJobDetail(\'' + run.job_id + '\')">[view job]</a>' : '')
+          + (run.job_id ? ' <a href="#" data-action="view-run-job" data-id="' + escapeHtml(run.job_id) + '">[view job]</a>' : '')
          + '</td>'
        + '<td>' + (run.tokens_used != null ? run.tokens_used : '-') + '</td>'
        + '</tr>';
@@ -3661,7 +3834,7 @@ function renderTeePopover(report) {
    + '<div class="tee-field"><div class="tee-field-label">VM Config</div>'
    + '<div class="tee-field-value">' + escapeHtml(vmConfig) + '</div></div>'
    + '<div class="tee-popover-actions">'
-    + '<button class="tee-btn-copy" onclick="copyTeeReport()">Copy Full Report</button></div>';
+    + '<button class="tee-btn-copy" data-action="copy-tee-report">Copy Full Report</button></div>';
 }

 function copyTeeReport() {
@@ -4143,3 +4316,94 @@ function formatDate(isoString) {
  const d = new Date(isoString);
  return d.toLocaleString();
 }
+
+// --- Event Listener Registration (CSP-safe, no inline handlers) ---
+
+document.getElementById('auth-connect-btn').addEventListener('click', () => authenticate());
+document.getElementById('restart-overlay').addEventListener('click', () => cancelRestart());
+document.getElementById('restart-close-btn').addEventListener('click', () => cancelRestart());
+document.getElementById('restart-cancel-btn').addEventListener('click', () => cancelRestart());
+document.getElementById('restart-confirm-btn').addEventListener('click', () => confirmRestart());
+document.getElementById('restart-btn').addEventListener('click', () => triggerRestart());
+document.getElementById('thread-new-btn').addEventListener('click', () => createNewThread());
+document.getElementById('thread-toggle-btn').addEventListener('click', () => toggleThreadSidebar());
+document.getElementById('assistant-thread').addEventListener('click', () => switchToAssistant());
+document.getElementById('send-btn').addEventListener('click', () => sendMessage());
+document.getElementById('memory-edit-btn').addEventListener('click', () => startMemoryEdit());
+document.getElementById('memory-save-btn').addEventListener('click', () => saveMemoryEdit());
+document.getElementById('memory-cancel-btn').addEventListener('click', () => cancelMemoryEdit());
+document.getElementById('logs-server-level').addEventListener('change', (e) => setServerLogLevel(e.target.value));
+document.getElementById('logs-pause-btn').addEventListener('click', () => toggleLogsPause());
+document.getElementById('logs-clear-btn').addEventListener('click', () => clearLogs());
+document.getElementById('wasm-install-btn').addEventListener('click', () => installWasmExtension());
+document.getElementById('mcp-add-btn').addEventListener('click', () => addMcpServer());
+document.getElementById('skill-search-btn').addEventListener('click', () => searchClawHub());
+document.getElementById('skill-install-btn').addEventListener('click', () => installSkillFromForm());
+
+// --- Delegated Event Handlers (for dynamically generated HTML) ---
+
+document.addEventListener('click', function(e) {
+  const el = e.target.closest('[data-action]');
+  if (!el) return;
+  const action = el.dataset.action;
+
+  switch (action) {
+    case 'copy-code':
+      copyCodeBlock(el);
+      break;
+    case 'breadcrumb-root':
+      e.preventDefault();
+      loadMemoryTree();
+      break;
+    case 'breadcrumb-file':
+      e.preventDefault();
+      readMemoryFile(el.dataset.path);
+      break;
+    case 'cancel-job':
+      e.stopPropagation();
+      cancelJob(el.dataset.id);
+      break;
+    case 'open-job':
+      openJobDetail(el.dataset.id);
+      break;
+    case 'close-job-detail':
+      closeJobDetail();
+      break;
+    case 'restart-job':
+      restartJob(el.dataset.id);
+      break;
+    case 'open-routine':
+      openRoutineDetail(el.dataset.id);
+      break;
+    case 'toggle-routine':
+      e.stopPropagation();
+      toggleRoutine(el.dataset.id);
+      break;
+    case 'trigger-routine':
+      e.stopPropagation();
+      triggerRoutine(el.dataset.id);
+      break;
+    case 'delete-routine':
+      e.stopPropagation();
+      deleteRoutine(el.dataset.id, el.dataset.name);
+      break;
+    case 'close-routine-detail':
+      closeRoutineDetail();
+      break;
+    case 'view-run-job':
+      e.preventDefault();
+      switchTab('jobs');
+      openJobDetail(el.dataset.id);
+      break;
+    case 'copy-tee-report':
+      copyTeeReport();
+      break;
+    case 'switch-language':
+      if (typeof switchLanguage === 'function') switchLanguage(el.dataset.lang);
+      break;
+  }
+});
+
+document.getElementById('language-btn').addEventListener('click', function() {
+  if (typeof toggleLanguageMenu === 'function') toggleLanguageMenu();
+});
@@ -9,12 +9,12 @@
  <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
  <link href="https://fonts.googleapis.com/css2?family=DM+Sans:wght@400;500;600;700&family=IBM+Plex+Mono:wght@400;500;600&display=swap" rel="stylesheet">
  <link rel="stylesheet" href="/style.css">
-  
+
  <!-- i18n Modules -->
  <script src="/i18n/index.js"></script>
  <script src="/i18n/en.js"></script>
  <script src="/i18n/zh-CN.js"></script>
-  
+
  <script
    src="https://cdnjs.cloudflare.com/ajax/libs/dompurify/3.2.3/purify.min.js"
    integrity="sha384-osZDKVu4ipZP703HmPOhWdyBajcFyjX2Psjk//TG1Rc0AdwEtuToaylrmcK3LdAl"
@@ -37,7 +37,7 @@
      <div class="auth-form">
        <label for="token-input" data-i18n="auth.tokenLabel">Gateway Token</label>
        <input type="password" id="token-input" data-i18n="auth.tokenPlaceholder" data-i18n-attr="placeholder" placeholder="Paste your auth token" autofocus>
-        <button onclick="authenticate()" data-i18n="auth.connect">Connect</button>
+        <button id="auth-connect-btn" data-i18n="auth.connect">Connect</button>
      </div>
      <div id="auth-error"></div>
      <p class="auth-hint" data-i18n="auth.hint">Enter the GATEWAY_AUTH_TOKEN from your .env configuration.</p>
@@ -46,11 +46,11 @@

  <!-- Restart Confirmation Modal -->
  <div id="restart-confirm-modal" class="restart-modal" style="display: none;">
-    <div class="restart-modal-overlay" onclick="cancelRestart()"></div>
+    <div class="restart-modal-overlay" id="restart-overlay"></div>
    <div class="restart-modal-content">
      <div class="restart-modal-header">
        <h2 data-i18n="restart.title">Restart IronClaw Instance</h2>
-        <button class="restart-modal-close" onclick="cancelRestart()" data-i18n="restart.closeTooltip" data-i18n-attr="title"
+        <button class="restart-modal-close" id="restart-close-btn" data-i18n="restart.closeTooltip" data-i18n-attr="title"
          title="Close">×</button>
      </div>
      <div class="restart-modal-body">
@@ -63,8 +63,8 @@
        </div>
      </div>
      <div class="restart-modal-footer">
-        <button class="restart-modal-btn cancel" onclick="cancelRestart()" data-i18n="restart.cancel">Cancel</button>
-        <button class="restart-modal-btn confirm" onclick="confirmRestart()" data-i18n="restart.confirm">Confirm Restart</button>
+        <button class="restart-modal-btn cancel" id="restart-cancel-btn" data-i18n="restart.cancel">Cancel</button>
+        <button class="restart-modal-btn confirm" id="restart-confirm-btn" data-i18n="restart.confirm">Confirm Restart</button>
      </div>
    </div>
  </div>
@@ -98,17 +98,17 @@
      <button data-tab="extensions" data-i18n="tab.extensions">Extensions</button>
      <button data-tab="skills" data-i18n="tab.skills">Skills</button>
      <div class="spacer"></div>
-      
+
      <!-- Language Switcher -->
      <div class="language-switcher">
-        <button class="language-btn" id="language-btn" type="button" onclick="toggleLanguageMenu()" title="Switch Language"
+        <button class="language-btn" id="language-btn" type="button" title="Switch Language"
          aria-label="Switch language" aria-haspopup="true" aria-expanded="false" aria-controls="language-menu">🌐</button>
        <div class="language-menu" id="language-menu" style="display: none;">
-          <button type="button" class="language-option" onclick="switchLanguage('en')" data-lang="en">English</button>
-          <button type="button" class="language-option" onclick="switchLanguage('zh-CN')" data-lang="zh-CN">简体中文</button>
+          <button type="button" class="language-option" data-action="switch-language" data-lang="en">English</button>
+          <button type="button" class="language-option" data-action="switch-language" data-lang="zh-CN">简体中文</button>
        </div>
      </div>
-      
+
      <button class="status-logs-btn" data-tab="logs" data-i18n="tab.logs" title="Logs">Logs</button>
      <div class="tee-shield" id="tee-shield" style="display:none" title="Running in a Trusted Execution Environment">
        <svg width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
@@ -122,7 +122,7 @@
        <span id="sse-status" data-i18n="status.connected">Connected</span>
        <div class="gateway-popover" id="gateway-popover"></div>
      </div>
-      <button class="restart-btn" id="restart-btn" onclick="triggerRestart()" data-i18n="status.restartTooltip"
+      <button class="restart-btn" id="restart-btn" data-i18n="status.restartTooltip"
        data-i18n-attr="title" title="Gracefully restart the process" style="display: none;">
        <svg id="restart-icon" width="13" height="13" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
          <path d="M23 4v6h-6"></path>
@@ -137,13 +137,13 @@
    <div class="tab-panel active" id="tab-chat">
      <div class="thread-sidebar" id="thread-sidebar">
        <div class="thread-sidebar-header">
-          <button class="thread-new-btn" onclick="createNewThread()" data-i18n="chat.newThread" data-i18n-attr="title"
+          <button class="thread-new-btn" id="thread-new-btn" data-i18n="chat.newThread" data-i18n-attr="title"
            title="New thread (Ctrl/Cmd+N)">+</button>
          <div class="spacer"></div>
-          <button class="thread-toggle-btn" id="thread-toggle-btn" onclick="toggleThreadSidebar()" data-i18n="chat.toggleSidebar"
+          <button class="thread-toggle-btn" id="thread-toggle-btn" data-i18n="chat.toggleSidebar"
            data-i18n-attr="title" title="Toggle sidebar">&laquo;</button>
        </div>
-        <div class="assistant-item" id="assistant-thread" onclick="switchToAssistant()">
+        <div class="assistant-item" id="assistant-thread">
          <span class="assistant-label" id="assistant-label" data-i18n="chat.assistant">Assistant</span>
          <span class="assistant-meta" id="assistant-meta"></span>
        </div>
@@ -161,7 +161,7 @@
          <input type="file" id="image-file-input" accept="image/*" multiple style="display:none">
          <button id="attach-btn" class="attach-btn" data-i18n="chat.attachImages" data-i18n-attr="title" title="Attach images"
            aria-label="Attach images">&#x1F4CE;</button>
-          <button id="send-btn" onclick="sendMessage()" data-i18n="chat.send">Send</button>
+          <button id="send-btn" data-i18n="chat.send">Send</button>
        </div>
      </div>
    </div>
@@ -178,7 +178,7 @@
        <div class="memory-content">
          <div class="memory-breadcrumb" id="memory-breadcrumb">
            <span id="memory-breadcrumb-path">workspace /</span>
-            <button class="memory-edit-btn" id="memory-edit-btn" style="display:none" onclick="startMemoryEdit()" data-i18n="memory.edit">Edit</button>
+            <button class="memory-edit-btn" id="memory-edit-btn" style="display:none" data-i18n="memory.edit">Edit</button>
          </div>
          <div class="memory-viewer" id="memory-viewer">
            <div class="empty" data-i18n="memory.selectFile">Select a file to view its contents</div>
@@ -186,8 +186,8 @@
          <div class="memory-editor" id="memory-editor" style="display:none">
            <textarea id="memory-edit-textarea"></textarea>
            <div class="memory-editor-actions">
-              <button class="btn-save" onclick="saveMemoryEdit()" data-i18n="memory.save">Save</button>
-              <button class="btn-cancel-edit" onclick="cancelMemoryEdit()" data-i18n="memory.cancel">Cancel</button>
+              <button class="btn-save" id="memory-save-btn" data-i18n="memory.save">Save</button>
+              <button class="btn-cancel-edit" id="memory-cancel-btn" data-i18n="memory.cancel">Cancel</button>
            </div>
          </div>
        </div>
@@ -219,7 +219,7 @@
    <div class="tab-panel" id="tab-logs">
      <div class="logs-container">
        <div class="logs-toolbar">
-          <select id="logs-server-level" onchange="setServerLogLevel(this.value)" title="Server-side log level (changes what the server emits)">
+          <select id="logs-server-level" title="Server-side log level (changes what the server emits)">
            <option value="error">Server: ERROR</option>
            <option value="warn">Server: WARN</option>
            <option value="info" selected>Server: INFO</option>
@@ -234,8 +234,8 @@
          </select>
          <input type="text" id="logs-target-filter" placeholder="Filter by target...">
          <label class="logs-checkbox"><input type="checkbox" id="logs-autoscroll" checked> <span data-i18n="logs.autoScroll">Auto-scroll</span></label>
-          <button id="logs-pause-btn" onclick="toggleLogsPause()" data-i18n="logs.pause">Pause</button>
-          <button onclick="clearLogs()" data-i18n="logs.clear">Clear</button>
+          <button id="logs-pause-btn" data-i18n="logs.pause">Pause</button>
+          <button id="logs-clear-btn" data-i18n="logs.clear">Clear</button>
        </div>
        <div class="logs-output" id="logs-output"></div>
      </div>
@@ -287,7 +287,7 @@
          <div class="ext-install-form">
            <input type="text" id="wasm-install-name" data-i18n-placeholder="common.name" placeholder="Extension name">
            <input type="text" id="wasm-install-url" placeholder="URL to .tar.gz bundle">
-            <button onclick="installWasmExtension()" data-i18n="extensions.install">Install</button>
+            <button id="wasm-install-btn" data-i18n="extensions.install">Install</button>
          </div>
        </div>
        <div class="extensions-section">
@@ -299,7 +299,7 @@
          <div class="ext-install-form">
            <input type="text" id="mcp-install-name" data-i18n-placeholder="common.name" placeholder="Server name">
            <input type="text" id="mcp-install-url" placeholder="MCP server URL (https://...)">
-            <button onclick="addMcpServer()" data-i18n="mcp.add">Add</button>
+            <button id="mcp-add-btn" data-i18n="mcp.add">Add</button>
          </div>
        </div>
        <div class="extensions-section">
@@ -320,7 +320,7 @@
          <h3 data-i18n="skills.searchClawHub">Search ClawHub</h3>
          <div class="skill-search-box">
            <input type="text" id="skill-search-input" data-i18n-placeholder="skills.searchPlaceholder" placeholder="Search...">
-            <button onclick="searchClawHub()" data-i18n="skills.search">Search</button>
+            <button id="skill-search-btn" data-i18n="skills.search">Search</button>
          </div>
          <div class="extensions-list" id="skill-search-results"></div>
        </div>
@@ -335,7 +335,7 @@
          <div class="ext-install-form">
            <input type="text" id="skill-install-name" data-i18n-placeholder="skills.namePlaceholder" placeholder="Skill name or slug">
            <input type="text" id="skill-install-url" data-i18n-placeholder="skills.urlPlaceholder" placeholder="HTTPS URL to SKILL.md (optional)">
-            <button onclick="installSkillFromForm()" data-i18n="extensions.install">Install</button>
+            <button id="skill-install-btn" data-i18n="extensions.install">Install</button>
          </div>
        </div>
      </div>
@@ -666,6 +666,7 @@ body {
  font-size: 14px;
  line-height: 1.5;
  word-wrap: break-word;
+  position: relative;
 }

 .message.user {
@@ -686,6 +687,58 @@ body {
  line-height: 1.6;
 }

+.message.has-copy {
+  padding-right: 52px;
+}
+
+.message-content {
+  min-width: 0;
+}
+
+.message-copy-btn {
+  position: absolute;
+  top: 8px;
+  right: 8px;
+  z-index: 2;
+  border: 1px solid var(--border);
+  background: var(--bg-primary);
+  color: var(--text-secondary);
+  border-radius: 8px;
+  font-size: 11px;
+  padding: 2px 8px;
+  opacity: 0;
+  pointer-events: none;
+  transition: opacity 0.15s ease;
+}
+
+.message.user:hover .message-copy-btn,
+.message.assistant:hover .message-copy-btn,
+.message.user:focus-within .message-copy-btn,
+.message.assistant:focus-within .message-copy-btn {
+  opacity: 1;
+  pointer-events: auto;
+}
+
+.message-copy-btn:focus-visible {
+  opacity: 1;
+  pointer-events: auto;
+  outline: 2px solid var(--accent);
+  outline-offset: 1px;
+}
+
+.message-copy-btn:hover {
+  background: var(--bg-secondary);
+  color: var(--text-primary);
+}
+
+@media (hover: none) {
+  .message.user .message-copy-btn,
+  .message.assistant .message-copy-btn {
+    opacity: 1;
+    pointer-events: auto;
+  }
+}
+
 .message.system {
  align-self: center;
  background: var(--bg-tertiary);
@@ -1166,7 +1219,21 @@ body {
  color: var(--danger);
 }

-/* Auth card (inline in chat) */
+/* Auth prompt */
+.auth-overlay {
+  position: fixed;
+  top: 0;
+  left: 0;
+  width: 100%;
+  height: 100%;
+  background: rgba(0, 0, 0, 0.6);
+  z-index: 1001;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  padding: 16px;
+}
+
 .auth-card {
  align-self: flex-start;
  max-width: 80%;
@@ -1181,6 +1248,16 @@ body {
  transition: border-color 0.2s;
 }

+.auth-overlay .auth-card {
+  width: 460px;
+  max-width: min(460px, 90vw);
+  margin: 0;
+  align-self: auto;
+  background: var(--bg);
+  border-color: rgba(52, 211, 153, 0.35);
+  box-shadow: 0 24px 48px rgba(0, 0, 0, 0.35);
+}
+
 .auth-card .auth-header {
  font-weight: 600;
  color: var(--accent);
@@ -3826,8 +3903,9 @@ mark {
  display: flex;
  flex-direction: row;
  gap: 8px;
-  padding: 4px;
+  padding: 10px 4px 4px;
  overflow-x: auto;
+  overflow-y: visible;
  min-height: 0;
  width: 100%;
 }
@@ -3848,6 +3926,7 @@ mark {
  border-radius: 6px;
  object-fit: cover;
  display: block;
+  border: 1px solid var(--border);
 }

 .image-preview-remove {
@@ -3856,21 +3935,41 @@ mark {
  right: -6px;
  width: 18px;
  height: 18px;
+  padding: 4px !important;
  border-radius: 50%;
  background: var(--danger);
  color: #fff;
-  border: none;
-  font-size: 12px;
-  line-height: 18px;
+  border: 2px solid var(--bg-secondary);
+  font-size: 10px;
+  line-height: 14px;
  text-align: center;
  cursor: pointer;
  padding: 0;
+  display: flex;
+  align-items: center;
+  justify-content: center;
 }

 .image-preview-remove:hover {
  background: #c33;
 }

+/* Images in user message bubbles */
+.message.user .message-images {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 6px;
+  margin-bottom: 4px;
+}
+
+.message.user .message-image-thumb {
+  max-width: 240px;
+  max-height: 200px;
+  border-radius: 8px;
+  object-fit: cover;
+  display: block;
+}
+
 /* Generated Image */
 .generated-image-card {
  max-width: 512px;
@@ -176,8 +176,12 @@ async fn handle_client_message(
                incoming = incoming.with_attachments(attachments);
            }

-            let tx_guard = state.msg_tx.read().await;
-            if let Some(ref tx) = *tx_guard {
+            // Clone sender to avoid holding RwLock read guard across send().await
+            let tx = {
+                let tx_guard = state.msg_tx.read().await;
+                tx_guard.as_ref().cloned()
+            };
+            if let Some(tx) = tx {
                if tx.send(incoming).await.is_err() {
                    let _ = direct_tx
                        .send(WsServerMessage::Error {
@@ -245,8 +249,12 @@ async fn handle_client_message(
            if let Some(ref tid) = thread_id {
                msg = msg.with_thread(tid);
            }
-            let tx_guard = state.msg_tx.read().await;
-            if let Some(ref tx) = *tx_guard {
+            // Clone sender to avoid holding RwLock read guard across send().await
+            let tx = {
+                let tx_guard = state.msg_tx.read().await;
+                tx_guard.as_ref().cloned()
+            };
+            if let Some(tx) = tx {
                let _ = tx.send(msg).await;
            }
        }
@@ -7,6 +7,7 @@
 //! - Managing WASM tools (`tool install`, `tool list`, `tool remove`)
 //! - Managing MCP servers (`mcp add`, `mcp auth`, `mcp list`, `mcp test`)
 //! - Querying workspace memory (`memory search`, `memory read`, `memory write`)
+//! - Managing routines (`routines list`, `routines create`, `routines edit`, ...)
 //! - Managing OS service (`service install`, `service start`, `service stop`)
 //! - Listing configured channels (`channels list`)
 //! - Active health diagnostics (`doctor`)
@@ -23,6 +24,7 @@ pub mod memory;
 pub mod oauth_defaults;
 mod pairing;
 mod registry;
+mod routines;
 mod service;
 mod skills;
 pub mod status;
@@ -39,6 +41,7 @@ pub use memory::MemoryCommand;
 pub use memory::run_memory_command_with_db;
 pub use pairing::{PairingCommand, run_pairing_command, run_pairing_command_with_store};
 pub use registry::{RegistryCommand, run_registry_command};
+pub use routines::{RoutinesCommand, run_routines_command};
 pub use service::{ServiceCommand, run_service_command};
 pub use skills::{SkillsCommand, run_skills_command};
 pub use status::run_status_command;
@@ -147,6 +150,15 @@ pub enum Command {
    )]
    Channels(ChannelsCommand),

+    /// Manage routines (scheduled, event-driven, webhook, manual)
+    #[command(
+        subcommand,
+        alias = "cron",
+        about = "Manage routines",
+        long_about = "List, create, edit, enable/disable, delete, and view history of routines.\nExamples:\n  ironclaw routines list\n  ironclaw routines create --name daily-digest --schedule '0 0 9 * * *' --prompt 'Summarize today'"
+    )]
+    Routines(RoutinesCommand),
+
    /// Manage MCP servers (hosted tool providers)
    #[command(
        subcommand,
@@ -281,6 +293,23 @@ pub async fn init_secrets_store()
    Ok(crate::db::create_secrets_store(&config.database, crypto).await?)
 }

+/// Run the Routines CLI subcommand.
+pub async fn run_routines_cli(
+    routines_cmd: &RoutinesCommand,
+    config_path: Option<&std::path::Path>,
+) -> anyhow::Result<()> {
+    let config = crate::config::Config::from_env_with_toml(config_path)
+        .await
+        .map_err(|e| anyhow::anyhow!("{e:#}"))?;
+
+    let db: Arc<dyn crate::db::Database> = crate::db::connect_from_config(&config.database)
+        .await
+        .map_err(|e| anyhow::anyhow!("{e:#}"))?;
+
+    let user_id = std::env::var("GATEWAY_USER_ID").unwrap_or_else(|_| "default".to_string());
+    run_routines_command(routines_cmd.clone(), db, &user_id).await
+}
+
 /// Run the Memory CLI subcommand.
 pub async fn run_memory_command(mem_cmd: &MemoryCommand) -> anyhow::Result<()> {
    let config = crate::config::Config::from_env()
@@ -172,6 +172,35 @@ pub async fn exchange_oauth_code(
    redirect_uri: &str,
    code_verifier: Option<&str>,
    access_token_field: &str,
+) -> Result<OAuthTokenResponse, OAuthCallbackError> {
+    // Delegates to exchange_oauth_code_with_resource with resource=None.
+    // Non-MCP OAuth flows don't need the RFC 8707 resource parameter.
+    exchange_oauth_code_with_resource(
+        token_url,
+        client_id,
+        client_secret,
+        code,
+        redirect_uri,
+        code_verifier,
+        access_token_field,
+        None,
+    )
+    .await
+}
+
+/// Exchange an OAuth authorization code for tokens, with optional RFC 8707 `resource` parameter.
+///
+/// The `resource` parameter scopes the issued token to a specific server (used by MCP OAuth).
+#[allow(clippy::too_many_arguments)]
+pub async fn exchange_oauth_code_with_resource(
+    token_url: &str,
+    client_id: &str,
+    client_secret: Option<&str>,
+    code: &str,
+    redirect_uri: &str,
+    code_verifier: Option<&str>,
+    access_token_field: &str,
+    resource: Option<&str>,
 ) -> Result<OAuthTokenResponse, OAuthCallbackError> {
    let client = reqwest::Client::new();
    let mut token_params = vec![
@@ -184,6 +213,12 @@ pub async fn exchange_oauth_code(
        token_params.push(("code_verifier", verifier.to_string()));
    }

+    // RFC 8707: include the `resource` parameter so the authorization server
+    // scopes the issued token to the specific MCP server (protected resource).
+    if let Some(resource) = resource {
+        token_params.push(("resource", resource.to_string()));
+    }
+
    let mut request = client.post(token_url);

    if let Some(secret) = client_secret {
@@ -388,6 +423,12 @@ pub struct PendingOAuthFlow {
    pub sse_sender: Option<tokio::sync::broadcast::Sender<crate::channels::web::types::SseEvent>>,
    /// Gateway auth token for authenticating with the platform token exchange proxy.
    pub gateway_token: Option<String>,
+    /// RFC 8707 resource parameter (MCP OAuth only).
+    /// Sent during token exchange to scope the token to a specific MCP server.
+    pub resource: Option<String>,
+    /// Secret name for persisting the client ID (MCP OAuth only).
+    /// Needed so token refresh can find the client_id after the session ends.
+    pub client_id_secret_name: Option<String>,
    /// When this flow was created (for expiry).
    pub created_at: std::time::Instant,
 }
@@ -975,4 +1016,42 @@ mod tests {
        assert_eq!(strip_instance_prefix("abc123"), "abc123");
        assert_eq!(strip_instance_prefix(""), "");
    }
+
+    /// Verify that `build_oauth_url` includes the RFC 8707 `resource` parameter
+    /// when passed through `extra_params`, which is how MCP OAuth gateway mode
+    /// scopes tokens to a specific MCP server.
+    #[test]
+    fn test_build_oauth_url_includes_resource_via_extra_params() {
+        use std::collections::HashMap;
+
+        use crate::cli::oauth_defaults::build_oauth_url;
+
+        let mut extra = HashMap::new();
+        extra.insert(
+            "resource".to_string(),
+            "https://mcp.example.com".to_string(),
+        );
+
+        let result = build_oauth_url(
+            "https://auth.example.com/authorize",
+            "client-123",
+            "https://gateway.example.com/oauth/callback",
+            &["read".to_string()],
+            true,
+            &extra,
+        );
+
+        // The resource parameter should be URL-encoded in the auth URL
+        assert!(
+            result
+                .url
+                .contains("resource=https%3A%2F%2Fmcp.example.com"),
+            "Expected resource param in URL: {}",
+            result.url
+        );
+        // State and PKCE should be present
+        assert!(result.url.contains("state="));
+        assert!(result.url.contains("code_challenge="));
+        assert!(result.code_verifier.is_some());
+    }
 }
@@ -0,0 +1,732 @@
+//! `ironclaw routines` — manage scheduled routines from the CLI.
+//!
+//! Provides subcommands for listing, creating, editing, enabling/disabling,
+//! deleting, and viewing run history of routines without starting the full agent.
+
+use std::sync::Arc;
+
+use chrono::{DateTime, Utc};
+use clap::Subcommand;
+use uuid::Uuid;
+
+use crate::agent::routine::{
+    NotifyConfig, Routine, RoutineAction, RoutineGuardrails, Trigger, next_cron_fire,
+};
+use crate::db::Database;
+
+/// Routines subcommands.
+#[derive(Subcommand, Debug, Clone)]
+pub enum RoutinesCommand {
+    /// List routines
+    List {
+        /// Filter by trigger type (e.g. "cron", "webhook", "event")
+        #[arg(long)]
+        trigger: Option<String>,
+
+        /// Include disabled routines
+        #[arg(long)]
+        disabled: bool,
+
+        /// Output as JSON (for scripting)
+        #[arg(long)]
+        json: bool,
+    },
+
+    /// Create a new cron routine
+    #[command(alias = "add")]
+    Create {
+        /// Routine name (must be unique per user)
+        #[arg(long)]
+        name: String,
+
+        /// Cron schedule (6-field: "sec min hour day month weekday")
+        #[arg(long)]
+        schedule: String,
+
+        /// Prompt for the LLM
+        #[arg(long)]
+        prompt: String,
+
+        /// Optional description
+        #[arg(long, default_value = "")]
+        description: String,
+
+        /// IANA timezone (e.g. "America/New_York")
+        #[arg(long)]
+        timezone: Option<String>,
+
+        /// Cooldown between fires in seconds
+        #[arg(long, default_value = "300")]
+        cooldown: u64,
+
+        /// Notification channel
+        #[arg(long)]
+        notify_channel: Option<String>,
+    },
+
+    /// Edit an existing routine
+    #[command(alias = "update")]
+    Edit {
+        /// Routine name
+        #[arg(long)]
+        name: String,
+
+        /// New schedule
+        #[arg(long)]
+        schedule: Option<String>,
+
+        /// New prompt
+        #[arg(long)]
+        prompt: Option<String>,
+
+        /// New description
+        #[arg(long)]
+        description: Option<String>,
+
+        /// New timezone
+        #[arg(long)]
+        timezone: Option<String>,
+
+        /// New cooldown in seconds
+        #[arg(long)]
+        cooldown: Option<u64>,
+    },
+
+    /// Enable a routine
+    Enable {
+        /// Routine name
+        name: String,
+    },
+
+    /// Disable a routine
+    Disable {
+        /// Routine name
+        name: String,
+    },
+
+    /// Delete a routine
+    #[command(alias = "rm")]
+    Delete {
+        /// Routine name
+        name: String,
+
+        /// Skip confirmation prompt
+        #[arg(short, long)]
+        yes: bool,
+    },
+
+    /// Show run history for a routine
+    #[command(alias = "runs")]
+    History {
+        /// Routine name
+        name: String,
+
+        /// Maximum number of runs to show
+        #[arg(short, long, default_value = "10")]
+        limit: i64,
+
+        /// Output as JSON (for scripting)
+        #[arg(long)]
+        json: bool,
+    },
+}
+
+/// Run a routines CLI command against the database.
+pub async fn run_routines_command(
+    cmd: RoutinesCommand,
+    db: Arc<dyn Database>,
+    user_id: &str,
+) -> anyhow::Result<()> {
+    match cmd {
+        RoutinesCommand::List {
+            trigger,
+            disabled,
+            json,
+        } => list(&db, user_id, trigger.as_deref(), disabled, json).await,
+        RoutinesCommand::Create {
+            name,
+            schedule,
+            prompt,
+            description,
+            timezone,
+            cooldown,
+            notify_channel,
+        } => {
+            create(
+                &db,
+                user_id,
+                &name,
+                &schedule,
+                &prompt,
+                &description,
+                timezone.as_deref(),
+                cooldown,
+                notify_channel,
+            )
+            .await
+        }
+        RoutinesCommand::Edit {
+            name,
+            schedule,
+            prompt,
+            description,
+            timezone,
+            cooldown,
+        } => {
+            edit(
+                &db,
+                user_id,
+                &name,
+                schedule.as_deref(),
+                prompt.as_deref(),
+                description.as_deref(),
+                timezone.as_deref(),
+                cooldown,
+            )
+            .await
+        }
+        RoutinesCommand::Enable { name } => set_enabled(&db, user_id, &name, true).await,
+        RoutinesCommand::Disable { name } => set_enabled(&db, user_id, &name, false).await,
+        RoutinesCommand::Delete { name, yes } => delete(&db, user_id, &name, yes).await,
+        RoutinesCommand::History { name, limit, json } => {
+            history(&db, user_id, &name, limit, json).await
+        }
+    }
+}
+
+// ── List ────────────────────────────────────────────────────
+
+async fn list(
+    db: &Arc<dyn Database>,
+    user_id: &str,
+    trigger_filter: Option<&str>,
+    show_disabled: bool,
+    json: bool,
+) -> anyhow::Result<()> {
+    let routines = db.list_routines(user_id).await?;
+
+    let filtered: Vec<&Routine> = routines
+        .iter()
+        .filter(|r| {
+            trigger_filter
+                .map(|t| r.trigger.type_tag() == t)
+                .unwrap_or(true)
+        })
+        .filter(|r| show_disabled || r.enabled)
+        .collect();
+
+    if json {
+        let items: Vec<serde_json::Value> = filtered
+            .iter()
+            .map(|r| {
+                serde_json::json!({
+                    "id": r.id.to_string(),
+                    "name": r.name,
+                    "trigger": r.trigger.type_tag(),
+                    "enabled": r.enabled,
+                    "next_fire_at": r.next_fire_at,
+                    "last_run_at": r.last_run_at,
+                    "run_count": r.run_count,
+                    "consecutive_failures": r.consecutive_failures,
+                })
+            })
+            .collect();
+        println!("{}", serde_json::to_string_pretty(&items)?);
+        return Ok(());
+    }
+
+    if filtered.is_empty() {
+        if let Some(t) = trigger_filter {
+            println!("No {t} routines found.");
+        } else {
+            println!("No routines found.");
+        }
+        return Ok(());
+    }
+
+    // Header
+    println!(
+        "{:<36}  {:<20}  {:<8}  {:<8}  {:<22}  {:<22}  {:>5}",
+        "ID", "NAME", "TRIGGER", "STATUS", "NEXT FIRE", "LAST RUN", "RUNS"
+    );
+    println!("{}", "-".repeat(130));
+
+    for r in &filtered {
+        let status = if r.enabled {
+            if r.consecutive_failures > 0 {
+                format!("err({})", r.consecutive_failures)
+            } else {
+                "active".to_string()
+            }
+        } else {
+            "disabled".to_string()
+        };
+
+        let next_fire = r
+            .next_fire_at
+            .map(format_relative)
+            .unwrap_or_else(|| "-".to_string());
+
+        let last_run = r
+            .last_run_at
+            .map(format_relative)
+            .unwrap_or_else(|| "-".to_string());
+
+        let name = truncate(&r.name, 20);
+
+        println!(
+            "{:<36}  {:<20}  {:<8}  {:<8}  {:<22}  {:<22}  {:>5}",
+            r.id,
+            name,
+            r.trigger.type_tag(),
+            status,
+            next_fire,
+            last_run,
+            r.run_count,
+        );
+    }
+
+    println!("\n{} routine(s)", filtered.len());
+    Ok(())
+}
+
+// ── Create ──────────────────────────────────────────────────
+
+#[allow(clippy::too_many_arguments)]
+async fn create(
+    db: &Arc<dyn Database>,
+    user_id: &str,
+    name: &str,
+    schedule: &str,
+    prompt: &str,
+    description: &str,
+    timezone: Option<&str>,
+    cooldown_secs: u64,
+    notify_channel: Option<String>,
+) -> anyhow::Result<()> {
+    validate_timezone_arg(timezone)?;
+
+    // Validate the cron expression by computing next fire.
+    let next_fire = next_cron_fire(schedule, timezone)
+        .map_err(|e| anyhow::anyhow!("Invalid cron schedule: {e}"))?;
+
+    // Check for name conflict.
+    if db.get_routine_by_name(user_id, name).await?.is_some() {
+        anyhow::bail!("Routine '{}' already exists", name);
+    }
+
+    let now = Utc::now();
+    let routine = Routine {
+        id: Uuid::new_v4(),
+        name: name.to_string(),
+        description: description.to_string(),
+        user_id: user_id.to_string(),
+        enabled: true,
+        trigger: Trigger::Cron {
+            schedule: schedule.to_string(),
+            timezone: timezone.map(String::from),
+        },
+        action: RoutineAction::Lightweight {
+            prompt: prompt.to_string(),
+            context_paths: Vec::new(),
+            max_tokens: 4096,
+            use_tools: false,
+            max_tool_rounds: 0,
+        },
+        guardrails: RoutineGuardrails {
+            cooldown: std::time::Duration::from_secs(cooldown_secs),
+            max_concurrent: 1,
+            dedup_window: None,
+        },
+        notify: NotifyConfig {
+            channel: notify_channel,
+            user: user_id.to_string(),
+            on_attention: true,
+            on_failure: true,
+            on_success: false,
+        },
+        last_run_at: None,
+        next_fire_at: next_fire,
+        run_count: 0,
+        consecutive_failures: 0,
+        state: serde_json::json!({}),
+        created_at: now,
+        updated_at: now,
+    };
+
+    db.create_routine(&routine).await?;
+
+    println!("Created routine '{}'", name);
+    println!("  ID:        {}", routine.id);
+    println!("  Schedule:  {}", schedule);
+    if let Some(tz) = timezone {
+        println!("  Timezone:  {}", tz);
+    }
+    if let Some(nf) = next_fire {
+        println!("  Next fire: {}", format_relative(nf));
+    }
+    Ok(())
+}
+
+// ── Edit ────────────────────────────────────────────────────
+
+#[allow(clippy::too_many_arguments)]
+async fn edit(
+    db: &Arc<dyn Database>,
+    user_id: &str,
+    name: &str,
+    schedule: Option<&str>,
+    prompt: Option<&str>,
+    description: Option<&str>,
+    timezone: Option<&str>,
+    cooldown: Option<u64>,
+) -> anyhow::Result<()> {
+    let mut routine = require_routine(db, user_id, name).await?;
+    validate_timezone_arg(timezone)?;
+
+    let mut changed = false;
+
+    // Update schedule if provided (only valid for cron routines).
+    if let Some(new_schedule) = schedule {
+        let tz = timezone.or(match &routine.trigger {
+            Trigger::Cron { timezone, .. } => timezone.as_deref(),
+            _ => None,
+        });
+        let next_fire = next_cron_fire(new_schedule, tz)
+            .map_err(|e| anyhow::anyhow!("Invalid cron schedule: {e}"))?;
+        routine.trigger = Trigger::Cron {
+            schedule: new_schedule.to_string(),
+            timezone: tz.map(String::from),
+        };
+        routine.next_fire_at = next_fire;
+        changed = true;
+    } else if let Some(tz) = timezone {
+        // Update only timezone, recompute next fire with existing schedule.
+        if let Trigger::Cron { ref schedule, .. } = routine.trigger {
+            let next_fire = next_cron_fire(schedule, Some(tz))
+                .map_err(|e| anyhow::anyhow!("Invalid cron schedule: {e}"))?;
+            routine.trigger = Trigger::Cron {
+                schedule: schedule.clone(),
+                timezone: Some(tz.to_string()),
+            };
+            routine.next_fire_at = next_fire;
+            changed = true;
+        } else {
+            anyhow::bail!("Cannot set timezone on non-cron trigger");
+        }
+    }
+
+    if let Some(new_prompt) = prompt {
+        match &mut routine.action {
+            RoutineAction::Lightweight { prompt: p, .. } => {
+                *p = new_prompt.to_string();
+                changed = true;
+            }
+            RoutineAction::FullJob { description: d, .. } => {
+                *d = new_prompt.to_string();
+                changed = true;
+            }
+        }
+    }
+
+    if let Some(new_desc) = description {
+        routine.description = new_desc.to_string();
+        changed = true;
+    }
+
+    if let Some(cd) = cooldown {
+        routine.guardrails.cooldown = std::time::Duration::from_secs(cd);
+        changed = true;
+    }
+
+    if !changed {
+        println!("No changes specified.");
+        return Ok(());
+    }
+
+    routine.updated_at = Utc::now();
+    db.update_routine(&routine).await?;
+    println!("Updated routine '{}'", name);
+    Ok(())
+}
+
+// ── Enable / Disable ────────────────────────────────────────
+
+async fn set_enabled(
+    db: &Arc<dyn Database>,
+    user_id: &str,
+    name: &str,
+    enabled: bool,
+) -> anyhow::Result<()> {
+    let mut routine = require_routine(db, user_id, name).await?;
+
+    if routine.enabled == enabled {
+        println!(
+            "Routine '{}' is already {}",
+            name,
+            if enabled { "enabled" } else { "disabled" }
+        );
+        return Ok(());
+    }
+
+    routine.enabled = enabled;
+
+    // Recompute next fire when enabling a cron routine.
+    if enabled
+        && let Trigger::Cron {
+            ref schedule,
+            ref timezone,
+        } = routine.trigger
+    {
+        routine.next_fire_at = next_cron_fire(schedule, timezone.as_deref())
+            .map_err(|e| anyhow::anyhow!("Failed to compute next fire for stored schedule: {e}"))?;
+    }
+
+    routine.updated_at = Utc::now();
+    db.update_routine(&routine).await?;
+    println!(
+        "{} routine '{}'",
+        if enabled { "Enabled" } else { "Disabled" },
+        name
+    );
+    Ok(())
+}
+
+// ── Delete ──────────────────────────────────────────────────
+
+async fn delete(
+    db: &Arc<dyn Database>,
+    user_id: &str,
+    name: &str,
+    skip_confirm: bool,
+) -> anyhow::Result<()> {
+    let routine = require_routine(db, user_id, name).await?;
+
+    if !skip_confirm {
+        println!("Routine: {}", routine.name);
+        println!("      ID: {}", routine.id);
+        println!(" Trigger: {}", routine.trigger.type_tag());
+        if let Trigger::Cron { ref schedule, .. } = routine.trigger {
+            println!("Schedule: {}", schedule);
+        }
+        println!("   Runs: {}", routine.run_count);
+        print!("\nDelete this routine? [y/N] ");
+        std::io::Write::flush(&mut std::io::stdout())?;
+
+        let mut input = String::new();
+        std::io::stdin().read_line(&mut input)?;
+        if !matches!(input.trim().to_lowercase().as_str(), "y" | "yes") {
+            println!("Cancelled.");
+            return Ok(());
+        }
+    }
+
+    let deleted = db.delete_routine(routine.id).await?;
+    if deleted {
+        println!("Deleted routine '{}'", name);
+    } else {
+        anyhow::bail!("Failed to delete routine '{}'", name);
+    }
+    Ok(())
+}
+
+// ── History ─────────────────────────────────────────────────
+
+async fn history(
+    db: &Arc<dyn Database>,
+    user_id: &str,
+    name: &str,
+    limit: i64,
+    json: bool,
+) -> anyhow::Result<()> {
+    let routine = require_routine(db, user_id, name).await?;
+
+    let limit = limit.clamp(1, 50);
+    let runs = db.list_routine_runs(routine.id, limit).await?;
+
+    if json {
+        let items: Vec<serde_json::Value> = runs
+            .iter()
+            .map(|run| {
+                serde_json::json!({
+                    "id": run.id.to_string(),
+                    "status": run.status.to_string(),
+                    "started_at": run.started_at,
+                    "completed_at": run.completed_at,
+                    "result_summary": run.result_summary,
+                    "tokens_used": run.tokens_used,
+                })
+            })
+            .collect();
+        println!("{}", serde_json::to_string_pretty(&items)?);
+        return Ok(());
+    }
+
+    if runs.is_empty() {
+        println!("No runs found for routine '{}'", name);
+        return Ok(());
+    }
+
+    println!("Run history for '{}' (last {}):\n", name, runs.len());
+
+    println!(
+        "{:<36}  {:<8}  {:<20}  {:<12}  SUMMARY",
+        "RUN ID", "STATUS", "STARTED", "DURATION"
+    );
+    println!("{}", "-".repeat(100));
+
+    for run in &runs {
+        let duration = run
+            .completed_at
+            .map(|end| {
+                let secs = (end - run.started_at).num_seconds();
+                if secs < 60 {
+                    format!("{}s", secs)
+                } else {
+                    format!("{}m{}s", secs / 60, secs % 60)
+                }
+            })
+            .unwrap_or_else(|| "running".to_string());
+
+        let summary = run
+            .result_summary
+            .as_deref()
+            .map(|s| truncate(s, 40))
+            .unwrap_or_else(|| "-".to_string());
+
+        println!(
+            "{:<36}  {:<8}  {:<20}  {:<12}  {}",
+            run.id,
+            run.status,
+            run.started_at.format("%Y-%m-%d %H:%M:%S"),
+            duration,
+            summary,
+        );
+    }
+
+    println!("\n{} run(s) shown", runs.len());
+    Ok(())
+}
+
+// ── Shared lookup ────────────────────────────────────────────
+
+/// Look up a routine by name.
+async fn require_routine(
+    db: &Arc<dyn Database>,
+    user_id: &str,
+    name: &str,
+) -> anyhow::Result<Routine> {
+    db.get_routine_by_name(user_id, name)
+        .await?
+        .ok_or_else(|| anyhow::anyhow!("Routine '{}' not found", name))
+}
+
+fn validate_timezone_arg(timezone: Option<&str>) -> anyhow::Result<()> {
+    if let Some(tz) = timezone
+        && crate::timezone::parse_timezone(tz).is_none()
+    {
+        anyhow::bail!("Invalid timezone: '{tz}' is not a valid IANA timezone");
+    }
+    Ok(())
+}
+
+// ── Helpers ─────────────────────────────────────────────────
+
+/// Format a datetime relative to now (e.g. "in 2h", "3m ago").
+fn format_relative(dt: DateTime<Utc>) -> String {
+    let now = Utc::now();
+    let diff = dt.signed_duration_since(now);
+    let secs = diff.num_seconds();
+
+    if secs.abs() < 60 {
+        if secs >= 0 {
+            "in <1m".to_string()
+        } else {
+            "<1m ago".to_string()
+        }
+    } else if secs.abs() < 3600 {
+        let mins = secs.abs() / 60;
+        if secs >= 0 {
+            format!("in {}m", mins)
+        } else {
+            format!("{}m ago", mins)
+        }
+    } else if secs.abs() < 86400 {
+        let hours = secs.abs() / 3600;
+        if secs >= 0 {
+            format!("in {}h", hours)
+        } else {
+            format!("{}h ago", hours)
+        }
+    } else {
+        let days = secs.abs() / 86400;
+        if secs >= 0 {
+            format!("in {}d", days)
+        } else {
+            format!("{}d ago", days)
+        }
+    }
+}
+
+/// Truncate a string to a maximum character length.
+fn truncate(s: &str, max_chars: usize) -> String {
+    if s.chars().count() <= max_chars {
+        s.to_string()
+    } else {
+        let truncated: String = s.chars().take(max_chars.saturating_sub(2)).collect();
+        format!("{}..", truncated)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn format_relative_future() {
+        let future = Utc::now() + chrono::Duration::hours(2);
+        let result = format_relative(future);
+        assert!(
+            result.starts_with("in "),
+            "expected 'in ...' for future time, got: {result}"
+        );
+    }
+
+    #[test]
+    fn format_relative_past() {
+        let past = Utc::now() - chrono::Duration::minutes(30);
+        let result = format_relative(past);
+        assert!(
+            result.ends_with(" ago"),
+            "expected '... ago' for past time, got: {result}"
+        );
+    }
+
+    #[test]
+    fn format_relative_days() {
+        let far_future = Utc::now() + chrono::Duration::days(3);
+        let result = format_relative(far_future);
+        assert!(result.contains('d'), "expected days in: {result}");
+    }
+
+    #[test]
+    fn truncate_short_string() {
+        assert_eq!(truncate("hello", 10), "hello");
+    }
+
+    #[test]
+    fn truncate_long_string() {
+        let result = truncate("hello world", 7);
+        assert_eq!(result, "hello..");
+    }
+
+    #[test]
+    fn truncate_multibyte_safe() {
+        // Ensure no panic on multi-byte characters.
+        let cjk = "你好世界测试";
+        let result = truncate(cjk, 4);
+        assert!(result.ends_with(".."), "got: {result}");
+        // Must be valid UTF-8 (would have panicked otherwise).
+        assert!(result.is_char_boundary(result.len()));
+    }
+}
@@ -1,33 +0,0 @@
---
-source: src/cli/mod.rs
-assertion_line: 302
-expression: help
---
-Secure personal AI assistant that protects your data and expands its capabilities
-
-Usage: ironclaw [OPTIONS] [COMMAND]
-
-Commands:
-  run         Run the AI agent
-  onboard     Run interactive setup wizard
-  config      Manage app configs
-  tool        Manage WASM tools
-  registry    Browse/install extensions
-  mcp         Manage MCP servers
-  memory      Manage workspace memory
-  pairing     Manage DM pairing
-  service     Manage OS service
-  doctor      Run diagnostics
-  status      Show system status
-  completion  Generate completions
-  import      Import from other AI systems
-  help        Print this message or the help of the given subcommand(s)
-
-Options:
-      --cli-only           Run in interactive CLI mode only (disable other channels)
-      --no-db              Skip database connection (for testing)
-  -m, --message <MESSAGE>  Single message mode - send one message and exit
-  -c, --config <CONFIG>    Configuration file path (optional, uses env vars by default)
-      --no-onboard         Skip first-run onboarding check
-  -h, --help               Print help (see more with '--help')
-  -V, --version            Print version
@@ -13,6 +13,7 @@ Commands:
  tool        Manage WASM tools
  registry    Browse/install extensions
  channels    Manage channels
+  routines    Manage routines
  mcp         Manage MCP servers
  memory      Manage workspace memory
  pairing     Manage DM pairing
@@ -1,49 +0,0 @@
---
-source: src/cli/mod.rs
-assertion_line: 318
-expression: help
---
-IronClaw is a secure AI assistant. Use 'ironclaw <subcommand> --help' for details.
-Examples:
-  ironclaw run  # Start the agent
-  ironclaw config list  # List configs
-
-Usage: ironclaw [OPTIONS] [COMMAND]
-
-Commands:
-  run         Run the AI agent
-  onboard     Run interactive setup wizard
-  config      Manage app configs
-  tool        Manage WASM tools
-  registry    Browse/install extensions
-  mcp         Manage MCP servers
-  memory      Manage workspace memory
-  pairing     Manage DM pairing
-  service     Manage OS service
-  doctor      Run diagnostics
-  status      Show system status
-  completion  Generate completions
-  import      Import from other AI systems
-  help        Print this message or the help of the given subcommand(s)
-
-Options:
-      --cli-only
-          Run in interactive CLI mode only (disable other channels)
-
-      --no-db
-          Skip database connection (for testing)
-
-  -m, --message <MESSAGE>
-          Single message mode - send one message and exit
-
-  -c, --config <CONFIG>
-          Configuration file path (optional, uses env vars by default)
-
-      --no-onboard
-          Skip first-run onboarding check
-
-  -h, --help
-          Print help (see a summary with '-h')
-
-  -V, --version
-          Print version
@@ -16,6 +16,7 @@ Commands:
  tool        Manage WASM tools
  registry    Browse/install extensions
  channels    Manage channels
+  routines    Manage routines
  mcp         Manage MCP servers
  memory      Manage workspace memory
  pairing     Manage DM pairing
@@ -23,6 +23,9 @@ pub struct EmbeddingsConfig {
    pub ollama_base_url: String,
    /// Embedding vector dimension. Inferred from the model name when not set explicitly.
    pub dimension: usize,
+    /// Custom base URL for OpenAI-compatible embedding providers.
+    /// When set, overrides the default `https://api.openai.com`.
+    pub openai_base_url: Option<String>,
 }

 impl Default for EmbeddingsConfig {
@@ -36,6 +39,7 @@ impl Default for EmbeddingsConfig {
            model,
            ollama_base_url: "http://localhost:11434".to_string(),
            dimension,
+            openai_base_url: None,
        }
    }
 }
@@ -74,6 +78,8 @@ impl EmbeddingsConfig {

        let enabled = parse_bool_env("EMBEDDING_ENABLED", settings.embeddings.enabled)?;

+        let openai_base_url = optional_env("EMBEDDING_BASE_URL")?;
+
        Ok(Self {
            enabled,
            provider,
@@ -81,6 +87,7 @@ impl EmbeddingsConfig {
            model,
            ollama_base_url,
            dimension,
+            openai_base_url,
        })
    }

@@ -130,16 +137,27 @@ impl EmbeddingsConfig {
            }
            _ => {
                if let Some(api_key) = self.openai_api_key() {
-                    tracing::debug!(
-                        "Embeddings enabled via OpenAI (model: {}, dim: {})",
-                        self.model,
-                        self.dimension,
-                    );
-                    Some(Arc::new(crate::workspace::OpenAiEmbeddings::with_model(
+                    let mut provider = crate::workspace::OpenAiEmbeddings::with_model(
                        api_key,
                        &self.model,
                        self.dimension,
-                    )))
+                    );
+                    if let Some(ref base_url) = self.openai_base_url {
+                        tracing::debug!(
+                            "Embeddings enabled via OpenAI (model: {}, base_url: {}, dim: {})",
+                            self.model,
+                            base_url,
+                            self.dimension,
+                        );
+                        provider = provider.with_base_url(base_url);
+                    } else {
+                        tracing::debug!(
+                            "Embeddings enabled via OpenAI (model: {}, dim: {})",
+                            self.model,
+                            self.dimension,
+                        );
+                    }
+                    Some(Arc::new(provider))
                } else {
                    tracing::warn!("Embeddings configured but OPENAI_API_KEY not set");
                    None
@@ -164,6 +182,7 @@ mod tests {
            std::env::remove_var("EMBEDDING_PROVIDER");
            std::env::remove_var("EMBEDDING_MODEL");
            std::env::remove_var("OPENAI_API_KEY");
+            std::env::remove_var("EMBEDDING_BASE_URL");
        }
    }

@@ -247,4 +266,41 @@ mod tests {
            std::env::remove_var("EMBEDDING_ENABLED");
        }
    }
+
+    #[test]
+    fn embedding_base_url_parsed_from_env() {
+        let _guard = ENV_MUTEX.lock().expect("env mutex poisoned");
+        clear_embedding_env();
+
+        // SAFETY: Under ENV_MUTEX, no concurrent env access.
+        unsafe {
+            std::env::set_var("EMBEDDING_BASE_URL", "https://custom.example.com");
+        }
+
+        let settings = Settings::default();
+        let config = EmbeddingsConfig::resolve(&settings).expect("resolve should succeed");
+        assert_eq!(
+            config.openai_base_url.as_deref(),
+            Some("https://custom.example.com"),
+            "EMBEDDING_BASE_URL env var should be parsed into openai_base_url"
+        );
+
+        // SAFETY: Under ENV_MUTEX.
+        unsafe {
+            std::env::remove_var("EMBEDDING_BASE_URL");
+        }
+    }
+
+    #[test]
+    fn embedding_base_url_defaults_to_none() {
+        let _guard = ENV_MUTEX.lock().expect("env mutex poisoned");
+        clear_embedding_env();
+
+        let settings = Settings::default();
+        let config = EmbeddingsConfig::resolve(&settings).expect("resolve should succeed");
+        assert!(
+            config.openai_base_url.is_none(),
+            "openai_base_url should be None when EMBEDDING_BASE_URL is not set"
+        );
+    }
 }
@@ -18,6 +18,7 @@ pub mod relay;
 mod routines;
 mod safety;
 mod sandbox;
+mod search;
 mod secrets;
 mod skills;
 mod transcription;
@@ -44,6 +45,7 @@ pub use self::routines::RoutineConfig;
 pub use self::safety::SafetyConfig;
 use self::safety::resolve_safety_config;
 pub use self::sandbox::{ClaudeCodeConfig, SandboxModeConfig};
+pub use self::search::WorkspaceSearchConfig;
 pub use self::secrets::SecretsConfig;
 pub use self::skills::SkillsConfig;
 pub use self::transcription::TranscriptionConfig;
@@ -91,6 +93,7 @@ pub struct Config {
    pub claude_code: ClaudeCodeConfig,
    pub skills: SkillsConfig,
    pub transcription: TranscriptionConfig,
+    pub search: WorkspaceSearchConfig,
    pub observability: crate::observability::ObservabilityConfig,
    /// Channel-relay integration (Slack via external relay service).
    /// Present only when both `CHANNEL_RELAY_URL` and `CHANNEL_RELAY_API_KEY` are set.
@@ -166,6 +169,7 @@ impl Config {
                ..SkillsConfig::default()
            },
            transcription: TranscriptionConfig::default(),
+            search: WorkspaceSearchConfig::default(),
            observability: crate::observability::ObservabilityConfig::default(),
            relay: None,
        }
@@ -318,6 +322,7 @@ impl Config {
            claude_code: ClaudeCodeConfig::resolve()?,
            skills: SkillsConfig::resolve()?,
            transcription: TranscriptionConfig::resolve(settings)?,
+            search: WorkspaceSearchConfig::resolve()?,
            observability: crate::observability::ObservabilityConfig {
                backend: std::env::var("OBSERVABILITY_BACKEND").unwrap_or_else(|_| "none".into()),
            },
@@ -0,0 +1,211 @@
+use crate::config::helpers::{optional_env, parse_optional_env};
+use crate::error::ConfigError;
+use crate::workspace::FusionStrategy;
+
+/// Workspace search configuration resolved from environment variables.
+#[derive(Debug, Clone)]
+pub struct WorkspaceSearchConfig {
+    /// Fusion strategy: "rrf" or "weighted".
+    pub fusion_strategy: FusionStrategy,
+    /// RRF constant k (default 60).
+    pub rrf_k: u32,
+    /// FTS weight for fusion.
+    ///
+    /// [`Default`] uses 0.5. When the configuration is resolved, per-strategy
+    /// defaults are applied: 0.5 (RRF) or 0.3 (weighted).
+    pub fts_weight: f32,
+    /// Vector weight for fusion.
+    ///
+    /// [`Default`] uses 0.5. When the configuration is resolved, per-strategy
+    /// defaults are applied: 0.5 (RRF) or 0.7 (weighted).
+    pub vector_weight: f32,
+}
+
+impl Default for WorkspaceSearchConfig {
+    fn default() -> Self {
+        Self {
+            fusion_strategy: FusionStrategy::default(),
+            rrf_k: 60,
+            fts_weight: 0.5,
+            vector_weight: 0.5,
+        }
+    }
+}
+
+impl WorkspaceSearchConfig {
+    pub(crate) fn resolve() -> Result<Self, ConfigError> {
+        let fusion_strategy = match optional_env("SEARCH_FUSION_STRATEGY")? {
+            Some(s) => match s.to_lowercase().as_str() {
+                "rrf" => FusionStrategy::Rrf,
+                "weighted" => FusionStrategy::WeightedScore,
+                other => {
+                    return Err(ConfigError::InvalidValue {
+                        key: "SEARCH_FUSION_STRATEGY".to_string(),
+                        message: format!("must be 'rrf' or 'weighted', got '{other}'"),
+                    });
+                }
+            },
+            None => FusionStrategy::default(),
+        };
+
+        let rrf_k = parse_optional_env("SEARCH_RRF_K", 60u32)?;
+
+        // Per-strategy weight defaults: RRF uses 0.5/0.5, weighted uses 0.3/0.7 (vector-biased).
+        let (default_fts, default_vec) = match fusion_strategy {
+            FusionStrategy::Rrf => (0.5f32, 0.5f32),
+            FusionStrategy::WeightedScore => (0.3f32, 0.7f32),
+        };
+        let fts_weight = parse_optional_env("SEARCH_FTS_WEIGHT", default_fts)?;
+        let vector_weight = parse_optional_env("SEARCH_VECTOR_WEIGHT", default_vec)?;
+
+        if !fts_weight.is_finite() || fts_weight < 0.0 {
+            return Err(ConfigError::InvalidValue {
+                key: "SEARCH_FTS_WEIGHT".to_string(),
+                message: "must be a finite, non-negative float".to_string(),
+            });
+        }
+        if !vector_weight.is_finite() || vector_weight < 0.0 {
+            return Err(ConfigError::InvalidValue {
+                key: "SEARCH_VECTOR_WEIGHT".to_string(),
+                message: "must be a finite, non-negative float".to_string(),
+            });
+        }
+        if matches!(fusion_strategy, FusionStrategy::WeightedScore)
+            && fts_weight == 0.0
+            && vector_weight == 0.0
+        {
+            return Err(ConfigError::InvalidValue {
+                key: "SEARCH_FTS_WEIGHT/SEARCH_VECTOR_WEIGHT".to_string(),
+                message: "weighted fusion requires at least one non-zero weight".to_string(),
+            });
+        }
+
+        Ok(Self {
+            fusion_strategy,
+            rrf_k,
+            fts_weight,
+            vector_weight,
+        })
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::config::helpers::ENV_MUTEX;
+
+    fn clear_search_env() {
+        // SAFETY: Only called under ENV_MUTEX in tests.
+        unsafe {
+            std::env::remove_var("SEARCH_FUSION_STRATEGY");
+            std::env::remove_var("SEARCH_RRF_K");
+            std::env::remove_var("SEARCH_FTS_WEIGHT");
+            std::env::remove_var("SEARCH_VECTOR_WEIGHT");
+        }
+    }
+
+    #[test]
+    fn defaults_when_no_env() {
+        let _guard = ENV_MUTEX.lock().expect("env mutex poisoned");
+        clear_search_env();
+
+        let config = WorkspaceSearchConfig::resolve().expect("should resolve");
+        assert_eq!(config.fusion_strategy, FusionStrategy::Rrf);
+        assert_eq!(config.rrf_k, 60);
+        assert!((config.fts_weight - 0.5).abs() < 0.001);
+        assert!((config.vector_weight - 0.5).abs() < 0.001);
+    }
+
+    #[test]
+    fn env_overrides() {
+        let _guard = ENV_MUTEX.lock().expect("env mutex poisoned");
+        clear_search_env();
+
+        // SAFETY: Under ENV_MUTEX.
+        unsafe {
+            std::env::set_var("SEARCH_FUSION_STRATEGY", "weighted");
+            std::env::set_var("SEARCH_RRF_K", "30");
+            std::env::set_var("SEARCH_FTS_WEIGHT", "0.9");
+            std::env::set_var("SEARCH_VECTOR_WEIGHT", "0.1");
+        }
+
+        let config = WorkspaceSearchConfig::resolve().expect("should resolve");
+        assert_eq!(config.fusion_strategy, FusionStrategy::WeightedScore);
+        assert_eq!(config.rrf_k, 30);
+        assert!((config.fts_weight - 0.9).abs() < 0.001);
+        assert!((config.vector_weight - 0.1).abs() < 0.001);
+
+        clear_search_env();
+    }
+
+    #[test]
+    fn invalid_strategy_rejected() {
+        let _guard = ENV_MUTEX.lock().expect("env mutex poisoned");
+        clear_search_env();
+
+        // SAFETY: Under ENV_MUTEX.
+        unsafe {
+            std::env::set_var("SEARCH_FUSION_STRATEGY", "bm25");
+        }
+
+        let result = WorkspaceSearchConfig::resolve();
+        assert!(result.is_err());
+
+        clear_search_env();
+    }
+
+    #[test]
+    fn weighted_strategy_defaults() {
+        let _guard = ENV_MUTEX.lock().expect("env mutex poisoned");
+        clear_search_env();
+
+        // SAFETY: Under ENV_MUTEX.
+        unsafe {
+            std::env::set_var("SEARCH_FUSION_STRATEGY", "weighted");
+        }
+
+        let config = WorkspaceSearchConfig::resolve().expect("should resolve");
+        assert_eq!(config.fusion_strategy, FusionStrategy::WeightedScore);
+        // Weighted mode should default to 0.3 FTS / 0.7 vector
+        assert!((config.fts_weight - 0.3).abs() < 0.001);
+        assert!((config.vector_weight - 0.7).abs() < 0.001);
+
+        clear_search_env();
+    }
+
+    #[test]
+    fn weighted_both_zero_rejected() {
+        let _guard = ENV_MUTEX.lock().expect("env mutex poisoned");
+        clear_search_env();
+
+        // SAFETY: Under ENV_MUTEX.
+        unsafe {
+            std::env::set_var("SEARCH_FUSION_STRATEGY", "weighted");
+            std::env::set_var("SEARCH_FTS_WEIGHT", "0.0");
+            std::env::set_var("SEARCH_VECTOR_WEIGHT", "0.0");
+        }
+
+        let result = WorkspaceSearchConfig::resolve();
+        assert!(result.is_err());
+
+        clear_search_env();
+    }
+
+    #[test]
+    fn rrf_both_zero_allowed() {
+        let _guard = ENV_MUTEX.lock().expect("env mutex poisoned");
+        clear_search_env();
+
+        // SAFETY: Under ENV_MUTEX.
+        unsafe {
+            std::env::set_var("SEARCH_FTS_WEIGHT", "0.0");
+            std::env::set_var("SEARCH_VECTOR_WEIGHT", "0.0");
+        }
+
+        // RRF ignores weights, so both=0 is fine
+        let config = WorkspaceSearchConfig::resolve().expect("should resolve");
+        assert_eq!(config.fusion_strategy, FusionStrategy::Rrf);
+
+        clear_search_env();
+    }
+}
@@ -16,6 +16,7 @@ mod workspace;

 use std::path::Path;
 use std::sync::Arc;
+use std::sync::atomic::{AtomicBool, Ordering};

 use async_trait::async_trait;
 use chrono::{DateTime, NaiveDateTime, Utc};
@@ -32,6 +33,8 @@ use crate::workspace::MemoryDocument;

 use crate::db::libsql_migrations;

+static NAIVE_TIMESTAMP_LOGGED: AtomicBool = AtomicBool::new(false);
+
 /// Explicit column list for routines table (matches positional access in `row_to_routine_libsql`).
 pub(crate) const ROUTINE_COLUMNS: &str = "\
    id, name, description, user_id, enabled, \
@@ -163,24 +166,27 @@ impl LibSqlBackend {
 ///
 /// Returns an error if none of the formats match.
 pub(crate) fn parse_timestamp(s: &str) -> Result<DateTime<Utc>, String> {
+    let log_naive_timestamp_once = || {
+        if !NAIVE_TIMESTAMP_LOGGED.swap(true, Ordering::Relaxed) {
+            tracing::debug!(
+                timestamp = %s,
+                "parsed naive timestamp without timezone; assuming UTC for backward compatibility"
+            );
+        }
+    };
+
    // RFC 3339 (our canonical write format)
    if let Ok(dt) = DateTime::parse_from_rfc3339(s) {
        return Ok(dt.with_timezone(&Utc));
    }
    // Naive with fractional seconds (legacy or SQLite datetime() output)
    if let Ok(ndt) = NaiveDateTime::parse_from_str(s, "%Y-%m-%d %H:%M:%S%.f") {
-        tracing::warn!(
-            timestamp = %s,
-            "parsed naive timestamp without timezone; assuming UTC for backward compatibility"
-        );
+        log_naive_timestamp_once();
        return Ok(ndt.and_utc());
    }
    // Naive without fractional seconds (legacy format)
    if let Ok(ndt) = NaiveDateTime::parse_from_str(s, "%Y-%m-%d %H:%M:%S") {
-        tracing::warn!(
-            timestamp = %s,
-            "parsed naive timestamp without timezone; assuming UTC for backward compatibility"
-        );
+        log_naive_timestamp_once();
        return Ok(ndt.and_utc());
    }
    Err(format!("unparseable timestamp: {:?}", s))
@@ -14,7 +14,7 @@ use crate::db::WorkspaceStore;
 use crate::error::WorkspaceError;
 use crate::workspace::{
    MemoryChunk, MemoryDocument, RankedResult, SearchConfig, SearchResult, WorkspaceEntry,
-    reciprocal_rank_fusion,
+    fuse_results,
 };

 use chrono::Utc;
@@ -614,6 +614,6 @@ impl WorkspaceStore for LibSqlBackend {
            );
        }

-        Ok(reciprocal_rank_fusion(fts_results, vector_results, config))
+        Ok(fuse_results(fts_results, vector_results, config))
    }
 }
@@ -27,7 +27,7 @@ use crate::secrets::{CreateSecretParams, SecretsStore};
 use crate::tools::ToolRegistry;
 use crate::tools::mcp::McpClient;
 use crate::tools::mcp::auth::{
-    PkceChallenge, authorize_mcp_server, build_authorization_url, discover_full_oauth_metadata,
+    authorize_mcp_server, canonical_resource_uri, discover_full_oauth_metadata,
    find_available_port, is_authenticated, register_client,
 };
 use crate::tools::mcp::config::McpServerConfig;
@@ -108,6 +108,13 @@ pub struct ExtensionManager {
    /// Relay config captured at startup. Used by `auth_channel_relay` and
    /// `activate_channel_relay` instead of re-reading env vars.
    relay_config: Option<crate::config::RelayConfig>,
+    /// When `true`, OAuth flows always return an auth URL to the caller
+    /// instead of opening a browser on the server via `open::that()`.
+    /// Set by the web gateway at startup via `enable_gateway_mode()`.
+    gateway_mode: std::sync::atomic::AtomicBool,
+    /// The gateway's own base URL for building OAuth redirect URIs.
+    /// Set by the web gateway at startup via `enable_gateway_mode()`.
+    gateway_base_url: RwLock<Option<String>>,
 }

 /// Sanitize a URL for logging by removing query parameters and credentials.
@@ -181,9 +188,75 @@ impl ExtensionManager {
            pending_oauth_flows: crate::cli::oauth_defaults::new_pending_oauth_registry(),
            gateway_token: std::env::var("GATEWAY_AUTH_TOKEN").ok(),
            relay_config: crate::config::RelayConfig::from_env(),
+            gateway_mode: std::sync::atomic::AtomicBool::new(false),
+            gateway_base_url: RwLock::new(None),
        }
    }

+    /// Enable gateway mode so OAuth flows return auth URLs to the frontend
+    /// instead of calling `open::that()` on the server.
+    ///
+    /// `base_url` is the gateway's own public URL (e.g. `https://my-gateway.example.com`),
+    /// used to build OAuth redirect URIs when `IRONCLAW_OAUTH_CALLBACK_URL` is not set.
+    pub async fn enable_gateway_mode(&self, base_url: String) {
+        self.gateway_mode
+            .store(true, std::sync::atomic::Ordering::Release);
+        *self.gateway_base_url.write().await = Some(base_url);
+    }
+
+    /// Returns `true` if OAuth should use gateway mode (return auth URL to
+    /// frontend) rather than CLI mode (open browser on server via `open::that`).
+    ///
+    /// Gateway mode is active when any of:
+    /// - `enable_gateway_mode()` was called (web gateway is running), OR
+    /// - `IRONCLAW_OAUTH_CALLBACK_URL` is set to a non-loopback URL, OR
+    /// - `self.tunnel_url` is set to a non-loopback URL
+    pub fn should_use_gateway_mode(&self) -> bool {
+        if self.gateway_mode.load(std::sync::atomic::Ordering::Acquire) {
+            return true;
+        }
+        if crate::cli::oauth_defaults::use_gateway_callback() {
+            return true;
+        }
+        self.tunnel_url
+            .as_ref()
+            .filter(|u| !u.is_empty())
+            .and_then(|raw| url::Url::parse(raw).ok())
+            .and_then(|u| u.host_str().map(String::from))
+            .map(|host| !crate::cli::oauth_defaults::is_loopback_host(&host))
+            .unwrap_or(false)
+    }
+
+    /// Returns the OAuth redirect URI for gateway mode, or `None` for local mode.
+    ///
+    /// Priority:
+    /// 1. `IRONCLAW_OAUTH_CALLBACK_URL` env var (via `callback_url()`)
+    /// 2. `gateway_base_url` (set by `enable_gateway_mode()`)
+    /// 3. `tunnel_url` (from config)
+    /// 4. `None` (local/CLI mode)
+    async fn gateway_callback_redirect_uri(&self) -> Option<String> {
+        use crate::cli::oauth_defaults;
+        if oauth_defaults::use_gateway_callback() {
+            return Some(format!("{}/oauth/callback", oauth_defaults::callback_url()));
+        }
+        // Use gateway_base_url from enable_gateway_mode()
+        if let Some(ref base) = *self.gateway_base_url.read().await {
+            let base = base.trim_end_matches('/');
+            return Some(format!("{}/oauth/callback", base));
+        }
+        // Fall back to tunnel_url
+        self.tunnel_url
+            .as_ref()
+            .filter(|u| !u.is_empty())
+            .and_then(|raw| url::Url::parse(raw).ok())
+            .and_then(|u| u.host_str().map(String::from))
+            .filter(|host| !oauth_defaults::is_loopback_host(host))
+            .map(|_| {
+                let base = self.tunnel_url.as_ref().unwrap().trim_end_matches('/');
+                format!("{}/oauth/callback", base)
+            })
+    }
+
    /// Get the relay config stored at startup.
    fn relay_config(&self) -> Result<&crate::config::RelayConfig, ExtensionError> {
        self.relay_config.as_ref().ok_or_else(|| {
@@ -193,6 +266,12 @@ impl ExtensionManager {
        })
    }

+    /// Inject a registry entry for testing. The entry is added to the discovery
+    /// cache so it appears in search results alongside built-in entries.
+    pub async fn inject_registry_entry(&self, entry: crate::extensions::RegistryEntry) {
+        self.registry.cache_discovered(vec![entry]).await;
+    }
+
    /// Configure the channel runtime infrastructure for hot-activating WASM channels.
    ///
    /// Call after construction (and after wrapping in `Arc`) once the channel
@@ -707,6 +786,19 @@ impl ExtensionManager {
        Self::validate_extension_name(name)?;
        let kind = self.determine_installed_kind(name).await?;

+        // Clean up any in-progress OAuth flows for this extension.
+        // TCP mode: abort the listener task so port 9876 is freed immediately.
+        // Gateway mode: remove stale pending flow entries.
+        if let Some(pending) = self.pending_auth.write().await.remove(name)
+            && let Some(handle) = pending.task_handle
+        {
+            handle.abort();
+        }
+        self.pending_oauth_flows
+            .write()
+            .await
+            .retain(|_, flow| flow.extension_name != name);
+
        match kind {
            ExtensionKind::McpServer => {
                // Unregister tools with this server's prefix
@@ -740,6 +832,14 @@ impl ExtensionManager {
                // Unregister from tool registry
                self.tool_registry.unregister(name).await;

+                // Evict compiled module from runtime cache so reinstall uses fresh binary
+                if let Some(ref rt) = self.wasm_tool_runtime {
+                    rt.remove(name).await;
+                }
+
+                // Clear stale activation errors so reinstall starts clean
+                self.activation_errors.write().await.remove(name);
+
                // Revoke credential mappings from the shared registry
                let cap_path = self
                    .wasm_tools_dir
@@ -780,6 +880,9 @@ impl ExtensionManager {
                self.active_channel_names.write().await.remove(name);
                self.persist_active_channels().await;

+                // Clear stale activation errors so reinstall starts clean
+                self.activation_errors.write().await.remove(name);
+
                // Delete channel files
                let wasm_path = self.wasm_channels_dir.join(format!("{}.wasm", name));
                let cap_path = self
@@ -1684,29 +1787,46 @@ impl ExtensionManager {
            return Ok(AuthResult::authenticated(name, ExtensionKind::McpServer));
        }

-        // Run the full OAuth flow (opens browser, waits for callback)
+        // In gateway mode, build an auth URL and return it for the frontend to
+        // open in the same browser. The gateway's /oauth/callback handler will
+        // complete the token exchange.
+        if self.should_use_gateway_mode() {
+            return match self.auth_mcp_build_url(name, &server).await {
+                Ok(result) => Ok(result),
+                Err(ExtensionError::AuthNotSupported(_)) => Ok(AuthResult::awaiting_token(
+                    name,
+                    ExtensionKind::McpServer,
+                    format!(
+                        "Server '{}' does not support OAuth. \
+                         Please provide an API token/key for this server.",
+                        name
+                    ),
+                    None,
+                )),
+                Err(e) => Err(e),
+            };
+        }
+
+        // CLI/local mode: run the full blocking OAuth flow (opens browser, waits for callback)
        match authorize_mcp_server(&server, &self.secrets, &self.user_id).await {
            Ok(_token) => {
                tracing::info!("MCP server '{}' authenticated via OAuth", name);
                Ok(AuthResult::authenticated(name, ExtensionKind::McpServer))
            }
            Err(crate::tools::mcp::auth::AuthError::NotSupported) => {
-                // Server doesn't support OAuth, try building a URL first
+                // Server doesn't support OAuth, try building a URL
                match self.auth_mcp_build_url(name, &server).await {
                    Ok(result) => Ok(result),
-                    Err(_) => {
-                        // No OAuth, no DCR: fall back to manual token entry
-                        Ok(AuthResult::awaiting_token(
-                            name,
-                            ExtensionKind::McpServer,
-                            format!(
-                                "Server '{}' does not support OAuth. \
-                                 Please provide an API token/key for this server.",
-                                name
-                            ),
-                            None,
-                        ))
-                    }
+                    Err(_) => Ok(AuthResult::awaiting_token(
+                        name,
+                        ExtensionKind::McpServer,
+                        format!(
+                            "Server '{}' does not support OAuth. \
+                             Please provide an API token/key for this server.",
+                            name
+                        ),
+                        None,
+                    )),
                }
            }
            Err(e) => {
@@ -1725,8 +1845,12 @@ impl ExtensionManager {
        }
    }

-    /// Build an auth URL for cases where non-interactive auth is needed
-    /// (e.g., running via Telegram where we can't open a browser).
+    /// Build an auth URL for MCP OAuth.
+    ///
+    /// In gateway mode, stores a `PendingOAuthFlow` so the web gateway's
+    /// `/oauth/callback` handler can complete the token exchange — the auth
+    /// URL is sent to the frontend which opens it in the same browser.
+    /// In local/CLI mode, builds the URL for the user to open manually.
    async fn auth_mcp_build_url(
        &self,
        name: &str,
@@ -1735,60 +1859,153 @@ impl ExtensionManager {
        // Try to discover OAuth metadata and build a URL the user can open manually
        let metadata = discover_full_oauth_metadata(&server.url)
            .await
-            .map_err(|e| ExtensionError::AuthFailed(e.to_string()))?;
+            .map_err(|e| match e {
+                crate::tools::mcp::auth::AuthError::NotSupported => {
+                    ExtensionError::AuthNotSupported(e.to_string())
+                }
+                _ => ExtensionError::AuthFailed(e.to_string()),
+            })?;
+
+        use crate::cli::oauth_defaults;
+
+        let is_gateway = self.should_use_gateway_mode();
+
+        // Build redirect URI: gateway uses the public callback URL,
+        // local mode binds a random port.
+        let redirect_uri = if let Some(uri) = self.gateway_callback_redirect_uri().await {
+            uri
+        } else {
+            let port = find_available_port()
+                .await
+                .map_err(|e| ExtensionError::AuthFailed(e.to_string()))?;
+            format!("http://localhost:{}/callback", port.1)
+        };

        // Try DCR if no client_id configured
-        let (client_id, redirect_uri) = if let Some(ref oauth) = server.oauth {
-            let port = find_available_port()
-                .await
-                .map_err(|e| ExtensionError::AuthFailed(e.to_string()))?;
-            let redirect = format!("http://localhost:{}/callback", port.1);
-            (oauth.client_id.clone(), redirect)
+        let (client_id, client_secret) = if let Some(ref oauth) = server.oauth {
+            (oauth.client_id.clone(), None)
        } else if let Some(ref reg_endpoint) = metadata.registration_endpoint {
-            let port = find_available_port()
-                .await
-                .map_err(|e| ExtensionError::AuthFailed(e.to_string()))?;
-            let redirect = format!("http://localhost:{}/callback", port.1);
-
-            let registration = register_client(reg_endpoint, &redirect)
+            let registration = register_client(reg_endpoint, &redirect_uri)
                .await
                .map_err(|e| ExtensionError::AuthFailed(e.to_string()))?;

-            (registration.client_id, redirect)
+            (registration.client_id, None)
        } else {
-            return Err(ExtensionError::AuthFailed(
+            return Err(ExtensionError::AuthNotSupported(
                "Server doesn't support OAuth or Dynamic Client Registration".to_string(),
            ));
        };

-        let pkce = PkceChallenge::generate();
-        let auth_url = build_authorization_url(
+        // RFC 8707: resource parameter to scope the token to this MCP server
+        let resource = canonical_resource_uri(&server.url);
+
+        // Build authorization URL with CSRF state using the shared oauth_defaults
+        // builder, which generates PKCE + state for us.
+        let mut extra_params = server
+            .oauth
+            .as_ref()
+            .map(|o| o.extra_params.clone())
+            .unwrap_or_default();
+        extra_params.insert("resource".to_string(), resource.clone());
+
+        let scopes = server
+            .oauth
+            .as_ref()
+            .map(|o| o.scopes.clone())
+            .unwrap_or_else(|| metadata.scopes_supported.clone());
+
+        let oauth_result = oauth_defaults::build_oauth_url(
            &metadata.authorization_endpoint,
            &client_id,
            &redirect_uri,
-            &metadata.scopes_supported,
-            Some(&pkce),
-            &std::collections::HashMap::new(),
-            None,
+            &scopes,
+            true, // Always use PKCE for MCP
+            &extra_params,
        );
+        let expected_state = oauth_result.state;
+        let code_verifier = oauth_result.code_verifier;

-        // Store pending auth for later callback handling
-        self.pending_auth.write().await.insert(
-            name.to_string(),
-            PendingAuth {
-                _name: name.to_string(),
-                _kind: ExtensionKind::McpServer,
+        if is_gateway {
+            // Gateway mode: store pending flow for the /oauth/callback handler.
+            oauth_defaults::sweep_expired_flows(&self.pending_oauth_flows).await;
+
+            // Platform routing: prepend instance name to state
+            let platform_state = oauth_defaults::build_platform_state(&expected_state);
+            let auth_url = if platform_state != expected_state {
+                oauth_result.url.replace(
+                    &format!("state={}", urlencoding::encode(&expected_state)),
+                    &format!("state={}", urlencoding::encode(&platform_state)),
+                )
+            } else {
+                oauth_result.url
+            };
+
+            let flow = oauth_defaults::PendingOAuthFlow {
+                extension_name: name.to_string(),
+                display_name: server.name.clone(),
+                token_url: metadata.token_endpoint,
+                client_id,
+                client_secret,
+                redirect_uri,
+                code_verifier,
+                access_token_field: "access_token".to_string(),
+                secret_name: server.token_secret_name(),
+                provider: Some(format!("mcp:{}", name)),
+                validation_endpoint: None,
+                scopes,
+                user_id: self.user_id.clone(),
+                secrets: Arc::clone(&self.secrets),
+                sse_sender: self.sse_sender.read().await.clone(),
+                gateway_token: self.gateway_token.clone(),
+                resource: Some(resource),
+                client_id_secret_name: if server.oauth.is_none() {
+                    Some(server.client_id_secret_name())
+                } else {
+                    None
+                },
                created_at: std::time::Instant::now(),
-                task_handle: None,
-            },
-        );
+            };

-        Ok(AuthResult::awaiting_authorization(
-            name,
-            ExtensionKind::McpServer,
-            auth_url,
-            "local".to_string(),
-        ))
+            self.pending_oauth_flows
+                .write()
+                .await
+                .insert(expected_state, flow);
+
+            self.pending_auth.write().await.insert(
+                name.to_string(),
+                PendingAuth {
+                    _name: name.to_string(),
+                    _kind: ExtensionKind::McpServer,
+                    created_at: std::time::Instant::now(),
+                    task_handle: None,
+                },
+            );
+
+            Ok(AuthResult::awaiting_authorization(
+                name,
+                ExtensionKind::McpServer,
+                auth_url,
+                "gateway".to_string(),
+            ))
+        } else {
+            // Local mode: return URL for manual opening
+            self.pending_auth.write().await.insert(
+                name.to_string(),
+                PendingAuth {
+                    _name: name.to_string(),
+                    _kind: ExtensionKind::McpServer,
+                    created_at: std::time::Instant::now(),
+                    task_handle: None,
+                },
+            );
+
+            Ok(AuthResult::awaiting_authorization(
+                name,
+                ExtensionKind::McpServer,
+                oauth_result.url,
+                "local".to_string(),
+            ))
+        }
    }

    async fn auth_wasm_tool(&self, name: &str) -> Result<AuthResult, ExtensionError> {
@@ -2203,7 +2420,10 @@ impl ExtensionManager {
            flows.retain(|_, flow| flow.extension_name != name);
        }

-        let redirect_uri = format!("{}/callback", oauth_defaults::callback_url());
+        let redirect_uri = self
+            .gateway_callback_redirect_uri()
+            .await
+            .unwrap_or_else(|| format!("{}/callback", oauth_defaults::callback_url()));

        // Merge scopes from all tools sharing this provider
        let merged_scopes = self
@@ -2228,7 +2448,7 @@ impl ExtensionManager {
            .clone()
            .unwrap_or_else(|| name.to_string());

-        if oauth_defaults::use_gateway_callback() {
+        if self.should_use_gateway_mode() {
            // Gateway mode: store pending flow state for the web gateway's
            // `/oauth/callback` handler to complete the exchange. No TCP listener
            // needed — the OAuth provider redirects to the gateway URL.
@@ -2264,6 +2484,8 @@ impl ExtensionManager {
                secrets: Arc::clone(&self.secrets),
                sse_sender: self.sse_sender.read().await.clone(),
                gateway_token: self.gateway_token.clone(),
+                resource: None,
+                client_id_secret_name: None,
                created_at: std::time::Instant::now(),
            };

@@ -2605,11 +2827,17 @@ impl ExtensionManager {
        .await
        .map_err(|e| ExtensionError::ActivationFailed(e.to_string()))?;

-        // Try to list and create tools
-        let mcp_tools = client
-            .list_tools()
-            .await
-            .map_err(|e| ExtensionError::ActivationFailed(e.to_string()))?;
+        // Try to list and create tools.
+        // A 401/auth error means the server requires OAuth — surface as
+        // AuthRequired so the activate handler triggers the OAuth flow.
+        let mcp_tools = client.list_tools().await.map_err(|e| {
+            let msg = e.to_string();
+            if msg.contains("requires authentication") || msg.contains("401") {
+                ExtensionError::AuthRequired
+            } else {
+                ExtensionError::ActivationFailed(msg)
+            }
+        })?;

        let tool_impls = client
            .create_tools()
@@ -2656,6 +2884,17 @@ impl ExtensionManager {
            });
        }

+        // Check auth status — block activation if required secrets are missing.
+        // NeedsAuth (OAuth not yet completed) is allowed because configure() loads
+        // the tool first, then starts the OAuth flow to obtain the token.
+        let auth_state = self.check_tool_auth_status(name).await;
+        if auth_state == ToolAuthState::NeedsSetup {
+            return Err(ExtensionError::ActivationFailed(format!(
+                "Tool '{}' requires configuration. Use the setup form to provide credentials.",
+                name
+            )));
+        }
+
        let runtime = self.wasm_tool_runtime.as_ref().ok_or_else(|| {
            ExtensionError::ActivationFailed("WASM runtime not available".to_string())
        })?;
@@ -4291,14 +4530,18 @@ mod tests {
    // available" because the ExtensionManager had `wasm_tool_runtime: None`.

    /// Build a minimal ExtensionManager suitable for unit tests.
-    fn make_test_manager(
+    fn make_test_manager_with_dirs(
        wasm_runtime: Option<Arc<crate::tools::wasm::WasmToolRuntime>>,
        tools_dir: std::path::PathBuf,
+        channels_dir: std::path::PathBuf,
    ) -> crate::extensions::manager::ExtensionManager {
        use crate::secrets::{InMemorySecretsStore, SecretsCrypto};
        use crate::tools::mcp::process::McpProcessManager;
        use crate::tools::mcp::session::McpSessionManager;

+        std::fs::create_dir_all(&tools_dir).ok();
+        std::fs::create_dir_all(&channels_dir).ok();
+
        let key = secrecy::SecretString::from(crate::secrets::keychain::generate_master_key_hex());
        let crypto = Arc::new(SecretsCrypto::new(key).expect("crypto"));
        let secrets: Arc<dyn crate::secrets::SecretsStore + Send + Sync> =
@@ -4313,15 +4556,22 @@ mod tests {
            tools,
            None, // hooks
            wasm_runtime,
-            tools_dir.clone(),
-            tools_dir, // channels dir (unused here)
-            None,      // tunnel_url
+            tools_dir,
+            channels_dir,
+            None, // tunnel_url
            "test".to_string(),
            None, // db
            vec![],
        )
    }

+    fn make_test_manager(
+        wasm_runtime: Option<Arc<crate::tools::wasm::WasmToolRuntime>>,
+        tools_dir: std::path::PathBuf,
+    ) -> crate::extensions::manager::ExtensionManager {
+        make_test_manager_with_dirs(wasm_runtime, tools_dir.clone(), tools_dir)
+    }
+
    #[tokio::test]
    async fn test_activate_wasm_tool_with_runtime_passes_runtime_check() {
        // When the ExtensionManager has a WASM runtime, activation should get
@@ -4674,6 +4924,145 @@ mod tests {
        );
    }

+    #[tokio::test]
+    async fn test_remove_wasm_tool_clears_pending_oauth_state_and_activation_error() {
+        let dir = tempfile::tempdir().expect("temp dir");
+        let mgr = make_test_manager(None, dir.path().to_path_buf());
+
+        std::fs::write(dir.path().join("gmail.wasm"), b"fake-tool").expect("write tool");
+
+        let listener = tokio::spawn(async {
+            std::future::pending::<()>().await;
+        });
+        let abort_handle = listener.abort_handle();
+        mgr.pending_auth.write().await.insert(
+            "gmail".to_string(),
+            super::PendingAuth {
+                _name: "gmail".to_string(),
+                _kind: ExtensionKind::WasmTool,
+                created_at: std::time::Instant::now(),
+                task_handle: Some(listener),
+            },
+        );
+
+        mgr.activation_errors
+            .write()
+            .await
+            .insert("gmail".to_string(), "cached failure".to_string());
+
+        let secrets = Arc::clone(&mgr.secrets);
+        mgr.pending_oauth_flows().write().await.insert(
+            "gmail-state".to_string(),
+            crate::cli::oauth_defaults::PendingOAuthFlow {
+                extension_name: "gmail".to_string(),
+                display_name: "Gmail".to_string(),
+                token_url: "https://example.com/token".to_string(),
+                client_id: "client123".to_string(),
+                client_secret: None,
+                redirect_uri: "https://example.com/oauth/callback".to_string(),
+                code_verifier: None,
+                access_token_field: "access_token".to_string(),
+                secret_name: "google_oauth_token".to_string(),
+                provider: None,
+                validation_endpoint: None,
+                scopes: vec![],
+                user_id: "test".to_string(),
+                secrets: Arc::clone(&secrets),
+                sse_sender: None,
+                gateway_token: None,
+                resource: None,
+                client_id_secret_name: None,
+                created_at: std::time::Instant::now(),
+            },
+        );
+        mgr.pending_oauth_flows().write().await.insert(
+            "other-state".to_string(),
+            crate::cli::oauth_defaults::PendingOAuthFlow {
+                extension_name: "web-search".to_string(),
+                display_name: "Web Search".to_string(),
+                token_url: "https://example.com/token".to_string(),
+                client_id: "client456".to_string(),
+                client_secret: None,
+                redirect_uri: "https://example.com/oauth/callback".to_string(),
+                code_verifier: None,
+                access_token_field: "access_token".to_string(),
+                secret_name: "other_token".to_string(),
+                provider: None,
+                validation_endpoint: None,
+                scopes: vec![],
+                user_id: "test".to_string(),
+                secrets,
+                sse_sender: None,
+                gateway_token: None,
+                resource: None,
+                client_id_secret_name: None,
+                created_at: std::time::Instant::now(),
+            },
+        );
+
+        let result = mgr.remove("gmail").await;
+        assert!(result.is_ok(), "remove should succeed: {:?}", result.err());
+
+        tokio::task::yield_now().await;
+
+        assert!(
+            mgr.pending_auth.read().await.get("gmail").is_none(),
+            "pending auth entry should be removed"
+        );
+        assert!(
+            abort_handle.is_finished(),
+            "pending auth listener should be aborted"
+        );
+        assert!(
+            !mgr.activation_errors.read().await.contains_key("gmail"),
+            "stale activation error should be cleared"
+        );
+
+        let flows = mgr.pending_oauth_flows().read().await;
+        assert!(
+            !flows.contains_key("gmail-state"),
+            "gateway OAuth flow for removed extension should be cleared"
+        );
+        assert!(
+            flows.contains_key("other-state"),
+            "unrelated pending OAuth flows should be retained"
+        );
+    }
+
+    #[tokio::test]
+    async fn test_remove_wasm_channel_clears_activation_error_and_deletes_files() {
+        let dir = tempfile::tempdir().expect("temp dir");
+        let tools_dir = dir.path().join("tools");
+        let channels_dir = dir.path().join("channels");
+        let mgr = make_test_manager_with_dirs(None, tools_dir, channels_dir.clone());
+
+        let wasm_path = channels_dir.join("telegram.wasm");
+        let cap_path = channels_dir.join("telegram.capabilities.json");
+        std::fs::write(&wasm_path, b"fake-channel").expect("write channel");
+        std::fs::write(&cap_path, b"{}").expect("write capabilities");
+
+        mgr.activation_errors
+            .write()
+            .await
+            .insert("telegram".to_string(), "channel failed".to_string());
+
+        let result = mgr.remove("telegram").await;
+        assert!(result.is_ok(), "remove should succeed: {:?}", result.err());
+
+        assert!(
+            !mgr.activation_errors.read().await.contains_key("telegram"),
+            "channel activation error should be cleared on remove"
+        );
+        assert!(
+            !wasm_path.exists(),
+            "channel wasm file should be deleted on remove"
+        );
+        assert!(
+            !cap_path.exists(),
+            "channel capabilities file should be deleted on remove"
+        );
+    }
+
    #[test]
    fn test_sanitize_url_with_query_params() {
        let url = "https://api.example.com/path?api_key=secret123&token=abc";
@@ -4766,6 +5155,189 @@ mod tests {
        assert!(result.contains("/v1/users/123/profile"));
    }

+    // ---- gateway mode detection tests ----
+    // Regression tests for a bug where MCP OAuth called `open::that()` on the
+    // server machine instead of returning an auth URL to the gateway frontend.
+    // The root cause was that `should_use_gateway_mode()` only checked the
+    // `IRONCLAW_OAUTH_CALLBACK_URL` env var, ignoring `self.tunnel_url`.
+
+    /// Serializes env-mutating tests to prevent parallel races.
+    static GATEWAY_ENV_MUTEX: std::sync::Mutex<()> = std::sync::Mutex::new(());
+
+    /// Build a minimal ExtensionManager with a custom tunnel_url.
+    fn make_manager_with_tunnel(tunnel_url: Option<String>) -> ExtensionManager {
+        use crate::secrets::{InMemorySecretsStore, SecretsCrypto};
+        use crate::tools::mcp::process::McpProcessManager;
+        use crate::tools::mcp::session::McpSessionManager;
+
+        let key = secrecy::SecretString::from(crate::secrets::keychain::generate_master_key_hex());
+        let crypto = Arc::new(SecretsCrypto::new(key).expect("crypto"));
+        let secrets: Arc<dyn crate::secrets::SecretsStore + Send + Sync> =
+            Arc::new(InMemorySecretsStore::new(crypto));
+        let tools = Arc::new(crate::tools::ToolRegistry::new());
+        let mcp = Arc::new(McpSessionManager::new());
+        let dir = std::env::temp_dir().join("ironclaw-test-gateway-mode");
+
+        ExtensionManager::new(
+            mcp,
+            Arc::new(McpProcessManager::new()),
+            secrets,
+            tools,
+            None,
+            None,
+            dir.clone(),
+            dir,
+            tunnel_url,
+            "test".to_string(),
+            None,
+            vec![],
+        )
+    }
+
+    #[test]
+    fn should_use_gateway_mode_true_for_tunnel_url() {
+        let _guard = GATEWAY_ENV_MUTEX.lock().expect("env mutex poisoned");
+        let original = std::env::var("IRONCLAW_OAUTH_CALLBACK_URL").ok();
+        // SAFETY: Under GATEWAY_ENV_MUTEX, no concurrent env access.
+        unsafe {
+            std::env::remove_var("IRONCLAW_OAUTH_CALLBACK_URL");
+        }
+
+        let mgr = make_manager_with_tunnel(Some("https://my-gateway.example.com".into()));
+        assert!(
+            mgr.should_use_gateway_mode(),
+            "should detect gateway mode from tunnel_url"
+        );
+
+        unsafe {
+            if let Some(val) = original {
+                std::env::set_var("IRONCLAW_OAUTH_CALLBACK_URL", val);
+            }
+        }
+    }
+
+    #[test]
+    fn should_use_gateway_mode_false_without_tunnel() {
+        let _guard = GATEWAY_ENV_MUTEX.lock().expect("env mutex poisoned");
+        let original = std::env::var("IRONCLAW_OAUTH_CALLBACK_URL").ok();
+        unsafe {
+            std::env::remove_var("IRONCLAW_OAUTH_CALLBACK_URL");
+        }
+
+        let mgr = make_manager_with_tunnel(None);
+        assert!(
+            !mgr.should_use_gateway_mode(),
+            "should not detect gateway mode without tunnel_url or env var"
+        );
+
+        unsafe {
+            if let Some(val) = original {
+                std::env::set_var("IRONCLAW_OAUTH_CALLBACK_URL", val);
+            }
+        }
+    }
+
+    #[test]
+    fn should_use_gateway_mode_false_for_loopback_tunnel() {
+        let _guard = GATEWAY_ENV_MUTEX.lock().expect("env mutex poisoned");
+        let original = std::env::var("IRONCLAW_OAUTH_CALLBACK_URL").ok();
+        unsafe {
+            std::env::remove_var("IRONCLAW_OAUTH_CALLBACK_URL");
+        }
+
+        let mgr = make_manager_with_tunnel(Some("http://127.0.0.1:3001".into()));
+        assert!(
+            !mgr.should_use_gateway_mode(),
+            "should not detect gateway mode for loopback tunnel_url"
+        );
+
+        unsafe {
+            if let Some(val) = original {
+                std::env::set_var("IRONCLAW_OAUTH_CALLBACK_URL", val);
+            }
+        }
+    }
+
+    /// Helper to run an async test body while holding the env mutex.
+    /// Clears `IRONCLAW_OAUTH_CALLBACK_URL` for the duration, restoring on drop.
+    struct EnvGuard {
+        original: Option<String>,
+        _mutex: std::sync::MutexGuard<'static, ()>,
+    }
+
+    impl EnvGuard {
+        fn new() -> Self {
+            let guard = GATEWAY_ENV_MUTEX.lock().expect("env mutex poisoned");
+            let original = std::env::var("IRONCLAW_OAUTH_CALLBACK_URL").ok();
+            // SAFETY: Under GATEWAY_ENV_MUTEX, no concurrent env access.
+            unsafe {
+                std::env::remove_var("IRONCLAW_OAUTH_CALLBACK_URL");
+            }
+            Self {
+                original,
+                _mutex: guard,
+            }
+        }
+    }
+
+    impl Drop for EnvGuard {
+        fn drop(&mut self) {
+            // SAFETY: Under GATEWAY_ENV_MUTEX (still held by _mutex), no concurrent env access.
+            unsafe {
+                if let Some(ref val) = self.original {
+                    std::env::set_var("IRONCLAW_OAUTH_CALLBACK_URL", val);
+                } else {
+                    std::env::remove_var("IRONCLAW_OAUTH_CALLBACK_URL");
+                }
+            }
+        }
+    }
+
+    #[tokio::test]
+    async fn gateway_callback_redirect_uri_from_tunnel_url() {
+        let _env = EnvGuard::new();
+
+        let mgr = make_manager_with_tunnel(Some("https://my-gateway.example.com".into()));
+        assert_eq!(
+            mgr.gateway_callback_redirect_uri().await,
+            Some("https://my-gateway.example.com/oauth/callback".to_string()),
+        );
+    }
+
+    #[tokio::test]
+    async fn gateway_callback_redirect_uri_none_without_tunnel() {
+        let _env = EnvGuard::new();
+
+        let mgr = make_manager_with_tunnel(None);
+        assert_eq!(mgr.gateway_callback_redirect_uri().await, None);
+    }
+
+    #[tokio::test]
+    async fn gateway_callback_redirect_uri_trims_trailing_slash() {
+        let _env = EnvGuard::new();
+
+        let mgr = make_manager_with_tunnel(Some("https://my-gateway.example.com/".into()));
+        assert_eq!(
+            mgr.gateway_callback_redirect_uri().await,
+            Some("https://my-gateway.example.com/oauth/callback".to_string()),
+        );
+    }
+
+    #[tokio::test]
+    async fn gateway_mode_enabled_explicitly() {
+        let _env = EnvGuard::new();
+
+        let mgr = make_manager_with_tunnel(None);
+        assert!(!mgr.should_use_gateway_mode());
+
+        mgr.enable_gateway_mode("https://my-gateway.example.com".into())
+            .await;
+        assert!(mgr.should_use_gateway_mode());
+        assert_eq!(
+            mgr.gateway_callback_redirect_uri().await,
+            Some("https://my-gateway.example.com/oauth/callback".to_string()),
+        );
+    }
    // ── Regression tests for PR #677 (unify-extension-lifecycle) ─────────

    #[tokio::test]
@@ -4915,7 +5487,6 @@ mod tests {
            "configure should have stored the relay stream token"
        );
    }
-
    #[test]
    fn test_validation_failed_is_distinct_error_variant() {
        // Regression: ValidationFailed must be a distinct error variant so
@@ -517,6 +517,9 @@ pub enum ExtensionError {
    #[error("Authentication failed: {0}")]
    AuthFailed(String),

+    #[error("Server does not support OAuth: {0}")]
+    AuthNotSupported(String),
+
    #[error("Activation failed: {0}")]
    ActivationFailed(String),

@@ -9,8 +9,8 @@ use ironclaw::{
    agent::{Agent, AgentDeps},
    app::{AppBuilder, AppBuilderFlags},
    channels::{
-        ChannelManager, ChannelSecretUpdater, GatewayChannel, HttpChannel, ReplChannel,
-        SignalChannel, WebhookServer, WebhookServerConfig,
+        ChannelManager, GatewayChannel, HttpChannel, ReplChannel, SignalChannel, WebhookServer,
+        WebhookServerConfig,
        wasm::{WasmChannelRouter, WasmChannelRuntime},
        web::log_layer::LogBroadcaster,
    },
@@ -67,6 +67,10 @@ async fn async_main() -> anyhow::Result<()> {
            )
            .await;
        }
+        Some(Command::Routines(routines_cmd)) => {
+            init_cli_tracing();
+            return ironclaw::cli::run_routines_cli(routines_cmd, cli.config.as_deref()).await;
+        }
        Some(Command::Mcp(mcp_cmd)) => {
            init_cli_tracing();
            return run_mcp_command(*mcp_cmd.clone()).await;
@@ -429,9 +433,8 @@ async fn async_main() -> anyhow::Result<()> {
        "Lifecycle hooks initialized"
    );

-    // Create session manager (shared between agent and web gateway)
-    let session_manager =
-        Arc::new(ironclaw::agent::SessionManager::new().with_hooks(components.hooks.clone()));
+    // Reuse the shared agent session manager prepared by AppBuilder.
+    let session_manager = Arc::clone(&components.agent_session_manager);

    // Lazy scheduler slot — filled after Agent::new creates the Scheduler.
    // Allows CreateJobTool to dispatch local jobs via the Scheduler even though
@@ -472,6 +475,14 @@ async fn async_main() -> anyhow::Result<()> {
        gw = gw.with_log_level_handle(Arc::clone(&log_level_handle));
        gw = gw.with_tool_registry(Arc::clone(&components.tools));
        if let Some(ref ext_mgr) = components.extension_manager {
+            // Enable gateway mode so MCP OAuth returns auth URLs to the frontend
+            // instead of calling open::that() on the server.
+            let gw_base = config
+                .tunnel
+                .public_url
+                .clone()
+                .unwrap_or_else(|| format!("http://{}:{}", gw_config.host, gw_config.port));
+            ext_mgr.enable_gateway_mode(gw_base).await;
            gw = gw.with_extension_manager(Arc::clone(ext_mgr));
        }
        if !components.catalog_entries.is_empty() {
@@ -725,6 +736,7 @@ async fn async_main() -> anyhow::Result<()> {

    #[cfg(unix)]
    {
+        use ironclaw::channels::ChannelSecretUpdater;
        // Collect all channels that support secret updates
        let mut secret_updaters: Vec<Arc<dyn ChannelSecretUpdater>> = Vec::new();
        if let Some(ref state) = http_channel_state {
@@ -65,7 +65,20 @@ fn install_macos() -> Result<()> {
    let stdout = logs_dir.join("daemon.stdout.log");
    let stderr = logs_dir.join("daemon.stderr.log");

-    let plist = format!(
+    let plist = macos_plist_content(
+        &exe.display().to_string(),
+        &stdout.display().to_string(),
+        &stderr.display().to_string(),
+    );
+
+    std::fs::write(&file, plist)?;
+    println!("Installed launchd service: {}", file.display());
+    println!("  Start with: ironclaw service start");
+    Ok(())
+}
+
+fn macos_plist_content(exe: &str, stdout: &str, stderr: &str) -> String {
+    format!(
        r#"<?xml version="1.0" encoding="UTF-8"?>
 <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
 <plist version="1.0">
@@ -81,6 +94,11 @@ fn install_macos() -> Result<()> {
  <true/>
  <key>KeepAlive</key>
  <true/>
+  <key>EnvironmentVariables</key>
+  <dict>
+    <key>CLI_ENABLED</key>
+    <string>false</string>
+  </dict>
  <key>StandardOutPath</key>
  <string>{stdout}</string>
  <key>StandardErrorPath</key>
@@ -89,15 +107,10 @@ fn install_macos() -> Result<()> {
 </plist>
 "#,
        label = SERVICE_LABEL,
-        exe = xml_escape(&exe.display().to_string()),
-        stdout = xml_escape(&stdout.display().to_string()),
-        stderr = xml_escape(&stderr.display().to_string()),
-    );
-
-    std::fs::write(&file, plist)?;
-    println!("Installed launchd service: {}", file.display());
-    println!("  Start with: ironclaw service start");
-    Ok(())
+        exe = xml_escape(exe),
+        stdout = xml_escape(stdout),
+        stderr = xml_escape(stderr),
+    )
 }

 fn install_linux() -> Result<()> {
@@ -356,4 +369,11 @@ mod tests {
        let s = path.to_string_lossy();
        assert!(s.ends_with(".ironclaw/logs"), "unexpected path: {s}");
    }
+
+    #[test]
+    fn macos_plist_sets_cli_enabled_false() {
+        let plist = macos_plist_content("/tmp/ironclaw", "/tmp/stdout.log", "/tmp/stderr.log");
+        assert!(plist.contains("<key>EnvironmentVariables</key>"));
+        assert!(plist.contains("    <key>CLI_ENABLED</key>\n    <string>false</string>"));
+    }
 }
@@ -1067,6 +1067,8 @@ mod tests {
                prompt: "Check status".to_string(),
                context_paths: vec![],
                max_tokens: 500,
+                use_tools: false,
+                max_tool_rounds: 3,
            },
            guardrails: RoutineGuardrails {
                cooldown: std::time::Duration::from_secs(60),
@@ -1198,6 +1200,8 @@ mod tests {
                prompt: "test".to_string(),
                context_paths: vec![],
                max_tokens: 100,
+                use_tools: false,
+                max_tool_rounds: 3,
            },
            guardrails: RoutineGuardrails {
                cooldown: std::time::Duration::from_secs(0),
@@ -256,7 +256,13 @@ impl Tool for ToolAuthTool {
    }

    fn requires_approval(&self, _params: &serde_json::Value) -> ApprovalRequirement {
-        ApprovalRequirement::UnlessAutoApproved
+        // In gateway mode, tool_auth only returns an auth URL for the frontend
+        // to open — no browser is launched server-side, so no approval needed.
+        if self.manager.should_use_gateway_mode() {
+            ApprovalRequirement::Never
+        } else {
+            ApprovalRequirement::UnlessAutoApproved
+        }
    }
 }

@@ -733,6 +739,22 @@ mod tests {
        }
    }

+    #[tokio::test]
+    async fn tool_auth_no_approval_in_gateway_mode() {
+        let manager = test_manager_stub();
+        manager
+            .enable_gateway_mode("http://localhost:3000".to_string())
+            .await;
+        let tool = ToolAuthTool {
+            manager: manager.clone(),
+        };
+        assert_eq!(
+            tool.requires_approval(&serde_json::json!({})),
+            ApprovalRequirement::Never,
+            "tool_auth should not require approval in gateway mode"
+        );
+    }
+
    #[test]
    fn test_tool_upgrade_schema() {
        use crate::tools::tool::ApprovalRequirement;
@@ -397,10 +397,6 @@ impl Tool for ListDirTool {
        false // Directory listings are safe
    }

-    fn requires_approval(&self, _params: &serde_json::Value) -> ApprovalRequirement {
-        ApprovalRequirement::UnlessAutoApproved
-    }
-
    fn domain(&self) -> ToolDomain {
        ToolDomain::Container
    }
@@ -31,6 +31,12 @@ const MAX_RESPONSE_SIZE: usize = 5 * 1024 * 1024;
 /// in memory for LLM context. Matches the WASM attachment size cap.
 const MAX_SAVE_TO_SIZE: usize = 50 * 1024 * 1024;

+/// Default request timeout when the caller does not provide one.
+const DEFAULT_TIMEOUT_SECS: u64 = 30;
+
+/// Maximum allowed request timeout to bound resource usage from LLM-controlled inputs.
+const MAX_TIMEOUT_SECS: u64 = 300;
+
 /// Maximum number of redirects to follow for simple GET requests.
 const MAX_REDIRECTS: usize = 3;

@@ -244,43 +250,120 @@ fn is_html_response(headers: &HashMap<String, String>) -> bool {
 fn parse_headers_param(
    headers: Option<&serde_json::Value>,
 ) -> Result<Vec<(String, String)>, ToolError> {
+    fn parse_header_object(
+        map: &serde_json::Map<String, serde_json::Value>,
+    ) -> Result<Vec<(String, String)>, ToolError> {
+        let mut out = Vec::with_capacity(map.len());
+        for (k, v) in map {
+            let value = v.as_str().ok_or_else(|| {
+                ToolError::InvalidParameters(format!("header '{}' must have a string value", k))
+            })?;
+            out.push((k.clone(), value.to_string()));
+        }
+        Ok(out)
+    }
+
+    fn parse_header_array(items: &[serde_json::Value]) -> Result<Vec<(String, String)>, ToolError> {
+        let mut out = Vec::with_capacity(items.len());
+        for (idx, item) in items.iter().enumerate() {
+            let obj = item.as_object().ok_or_else(|| {
+                ToolError::InvalidParameters(format!(
+                    "headers[{}] must be an object with 'name' and 'value'",
+                    idx
+                ))
+            })?;
+            let name = obj.get("name").and_then(|v| v.as_str()).ok_or_else(|| {
+                ToolError::InvalidParameters(format!("headers[{}].name must be a string", idx))
+            })?;
+            let value = obj.get("value").and_then(|v| v.as_str()).ok_or_else(|| {
+                ToolError::InvalidParameters(format!("headers[{}].value must be a string", idx))
+            })?;
+            out.push((name.to_string(), value.to_string()));
+        }
+        Ok(out)
+    }
+
    match headers {
        None => Ok(Vec::new()),
-        Some(serde_json::Value::Object(map)) => {
-            let mut out = Vec::with_capacity(map.len());
-            for (k, v) in map {
-                let value = v.as_str().ok_or_else(|| {
-                    ToolError::InvalidParameters(format!("header '{}' must have a string value", k))
-                })?;
-                out.push((k.clone(), value.to_string()));
+        Some(serde_json::Value::String(raw)) => {
+            let trimmed = raw.trim();
+            if trimmed.is_empty() {
+                return Ok(Vec::new());
            }
-            Ok(out)
-        }
-        Some(serde_json::Value::Array(items)) => {
-            let mut out = Vec::with_capacity(items.len());
-            for (idx, item) in items.iter().enumerate() {
-                let obj = item.as_object().ok_or_else(|| {
-                    ToolError::InvalidParameters(format!(
-                        "headers[{}] must be an object with 'name' and 'value'",
-                        idx
-                    ))
-                })?;
-                let name = obj.get("name").and_then(|v| v.as_str()).ok_or_else(|| {
-                    ToolError::InvalidParameters(format!("headers[{}].name must be a string", idx))
-                })?;
-                let value = obj.get("value").and_then(|v| v.as_str()).ok_or_else(|| {
-                    ToolError::InvalidParameters(format!("headers[{}].value must be a string", idx))
-                })?;
-                out.push((name.to_string(), value.to_string()));
+            let parsed = serde_json::from_str::<serde_json::Value>(trimmed).map_err(|e| {
+                ToolError::InvalidParameters(format!(
+                    "headers string must contain valid JSON object/array: {}",
+                    e
+                ))
+            })?;
+            match parsed {
+                serde_json::Value::Object(map) => parse_header_object(&map),
+                serde_json::Value::Array(items) => parse_header_array(&items),
+                _ => Err(ToolError::InvalidParameters(
+                    "headers string must decode to a JSON object or array".to_string(),
+                )),
            }
-            Ok(out)
        }
+        Some(serde_json::Value::Object(map)) => parse_header_object(map),
+        Some(serde_json::Value::Array(items)) => parse_header_array(items),
        Some(_) => Err(ToolError::InvalidParameters(
            "'headers' must be an object or an array of {name, value}".to_string(),
        )),
    }
 }

+fn parse_timeout_secs_param(timeout: Option<&serde_json::Value>) -> Result<Option<u64>, ToolError> {
+    let parsed = match timeout {
+        None | Some(serde_json::Value::Null) => Ok(None),
+        Some(serde_json::Value::Number(n)) => n.as_u64().map(Some).ok_or_else(|| {
+            ToolError::InvalidParameters("timeout_secs must be a non-negative integer".to_string())
+        }),
+        Some(serde_json::Value::String(raw)) => {
+            let trimmed = raw.trim();
+            if trimmed.is_empty() {
+                return Ok(None);
+            }
+            let secs = trimmed.parse::<u64>().map_err(|_| {
+                ToolError::InvalidParameters(
+                    "timeout_secs string must contain a non-negative integer".to_string(),
+                )
+            })?;
+            Ok(Some(secs))
+        }
+        Some(_) => Err(ToolError::InvalidParameters(
+            "timeout_secs must be an integer".to_string(),
+        )),
+    }?;
+
+    if let Some(secs) = parsed
+        && secs > MAX_TIMEOUT_SECS
+    {
+        return Err(ToolError::InvalidParameters(format!(
+            "timeout_secs must be <= {}",
+            MAX_TIMEOUT_SECS
+        )));
+    }
+
+    Ok(parsed)
+}
+
+fn parse_save_to_param(save_to: Option<&serde_json::Value>) -> Result<Option<String>, ToolError> {
+    match save_to {
+        None | Some(serde_json::Value::Null) => Ok(None),
+        Some(serde_json::Value::String(path)) => {
+            let trimmed = path.trim();
+            if trimmed.is_empty() {
+                Ok(None)
+            } else {
+                Ok(Some(trimmed.to_string()))
+            }
+        }
+        Some(_) => Err(ToolError::InvalidParameters(
+            "save_to must be a string".to_string(),
+        )),
+    }
+}
+
 /// Extract host from URL in params (for approval checks).
 fn extract_host_from_params(params: &serde_json::Value) -> Option<String> {
    params
@@ -315,7 +398,7 @@ impl Tool for HttpTool {
                "method": {
                    "type": "string",
                    "enum": ["GET", "POST", "PUT", "DELETE", "PATCH"],
-                    "description": "HTTP method"
+                    "description": "HTTP method (default: GET)"
                },
                "url": {
                    "type": "string",
@@ -346,7 +429,7 @@ impl Tool for HttpTool {
                    "description": "Save response body as raw bytes to this file path instead of returning it. Use for binary downloads (images, PDFs, etc.). The path must be under /tmp/."
                }
            },
-            "required": ["method", "url"]
+            "required": ["url"]
        })
    }

@@ -357,7 +440,8 @@ impl Tool for HttpTool {
    ) -> Result<ToolOutput, ToolError> {
        let start = std::time::Instant::now();

-        let method = require_str(&params, "method")?;
+        let method = params["method"].as_str().unwrap_or("GET");
+        let method_upper = method.to_uppercase();

        let url = require_str(&params, "url")?;
        let mut parsed_url = validate_url(url)?;
@@ -379,6 +463,9 @@ impl Tool for HttpTool {

        // Parse headers
        let mut headers_vec = parse_headers_param(params.get("headers"))?;
+        let timeout_secs = parse_timeout_secs_param(params.get("timeout_secs"))?;
+        let save_to = parse_save_to_param(params.get("save_to"))?;
+        let effective_timeout = Duration::from_secs(timeout_secs.unwrap_or(DEFAULT_TIMEOUT_SECS));

        // Build request
        let mut request = match method.to_uppercase().as_str() {
@@ -395,6 +482,8 @@ impl Tool for HttpTool {
            }
        };

+        request = request.timeout(effective_timeout);
+
        // Add headers
        for (key, value) in &headers_vec {
            request = request.header(key.as_str(), value.as_str());
@@ -403,7 +492,9 @@ impl Tool for HttpTool {
        // Add body if present
        let body_bytes = if let Some(body) = params.get("body") {
            if let Some(body_str) = body.as_str() {
-                if let Ok(json_body) = serde_json::from_str::<serde_json::Value>(body_str) {
+                if body_str.is_empty() {
+                    None
+                } else if let Ok(json_body) = serde_json::from_str::<serde_json::Value>(body_str) {
                    let bytes = serde_json::to_vec(&json_body).map_err(|e| {
                        ToolError::InvalidParameters(format!("invalid body JSON: {}", e))
                    })?;
@@ -468,7 +559,7 @@ impl Tool for HttpTool {

        // Build the interceptor request descriptor for recording/replay
        let intercept_req = crate::llm::recording::HttpExchangeRequest {
-            method: method.to_uppercase(),
+            method: method_upper,
            url: parsed_url.to_string(),
            headers: headers_vec.clone(),
            body: body_bytes
@@ -510,7 +601,7 @@ impl Tool for HttpTool {
                let hop_client = build_pinned_client(
                    &hop_host,
                    &hop_addrs,
-                    Duration::from_secs(30),
+                    effective_timeout,
                    reqwest::redirect::Policy::none(),
                )?;

@@ -524,7 +615,7 @@ impl Tool for HttpTool {
                    .await
                    .map_err(|e| {
                        if e.is_timeout() {
-                            ToolError::Timeout(Duration::from_secs(30))
+                            ToolError::Timeout(effective_timeout)
                        } else {
                            ToolError::ExternalService(e.to_string())
                        }
@@ -588,7 +679,7 @@ impl Tool for HttpTool {
        } else {
            let resp = request.send().await.map_err(|e| {
                if e.is_timeout() {
-                    ToolError::Timeout(Duration::from_secs(30))
+                    ToolError::Timeout(effective_timeout)
                } else {
                    ToolError::ExternalService(e.to_string())
                }
@@ -616,7 +707,7 @@ impl Tool for HttpTool {
            .collect();

        // Use a larger size limit when saving to disk (file downloads)
-        let saving_to_disk = params.get("save_to").is_some();
+        let saving_to_disk = save_to.is_some();
        let max_size = if saving_to_disk {
            MAX_SAVE_TO_SIZE
        } else {
@@ -661,11 +752,11 @@ impl Tool for HttpTool {
        let body_bytes = bytes::Bytes::from(body);

        // If save_to is specified, write raw bytes to file and return metadata.
-        if let Some(save_to) = params.get("save_to").and_then(|v| v.as_str()) {
-            let save_to_owned = save_to.to_string();
+        if let Some(save_to) = save_to {
+            let saved_to = save_to.clone();
            let bytes_clone = body_bytes.clone();
            tokio::task::spawn_blocking(move || {
-                let canonical = validate_save_to_path(&save_to_owned)?;
+                let canonical = validate_save_to_path(&save_to)?;
                std::fs::write(&canonical, &bytes_clone).map_err(|e| {
                    ToolError::ExecutionFailed(format!("failed to write file: {}", e))
                })?;
@@ -676,7 +767,7 @@ impl Tool for HttpTool {
            .map_err(|e: ToolError| e)?;
            let result = serde_json::json!({
                "status": status,
-                "saved_to": save_to,
+                "saved_to": saved_to,
                "size_bytes": body_bytes.len(),
                "headers": headers,
            });
@@ -738,18 +829,22 @@ impl Tool for HttpTool {
    }

    fn requires_approval(&self, params: &serde_json::Value) -> ApprovalRequirement {
-        // 1. Manual auth headers/query params in LLM params
-        if crate::safety::params_contain_manual_credentials(params) {
+        let has_credentials = crate::safety::params_contain_manual_credentials(params)
+            || (self.credential_registry.as_ref().is_some_and(|registry| {
+                extract_host_from_params(params)
+                    .is_some_and(|host| registry.has_credentials_for_host(&host))
+            }));
+
+        if has_credentials {
            return ApprovalRequirement::Always;
        }
-        // 2. Target host has credential mappings (will be auto-injected)
-        if let Some(ref registry) = self.credential_registry
-            && let Some(host) = extract_host_from_params(params)
-            && registry.has_credentials_for_host(&host)
-        {
-            return ApprovalRequirement::Always;
+
+        // GET requests (or missing method, since GET is the default) are low-risk
+        let method = params["method"].as_str().unwrap_or("GET");
+        if method.eq_ignore_ascii_case("GET") {
+            return ApprovalRequirement::Never;
        }
-        // Default: outbound HTTP still needs approval unless auto-approved
+
        ApprovalRequirement::UnlessAutoApproved
    }

@@ -887,6 +982,71 @@ mod tests {
        );
    }

+    #[test]
+    fn test_parse_headers_param_accepts_stringified_array() {
+        let headers =
+            serde_json::json!("[{\"name\":\"Authorization\",\"value\":\"Bearer token\"}]");
+        let parsed = parse_headers_param(Some(&headers)).unwrap();
+        assert_eq!(
+            parsed,
+            vec![("Authorization".to_string(), "Bearer token".to_string())]
+        );
+    }
+
+    #[test]
+    fn test_parse_headers_param_rejects_double_string_encoding() {
+        let headers = serde_json::json!("\"hello\"");
+        let err = parse_headers_param(Some(&headers)).unwrap_err();
+        assert!(
+            err.to_string()
+                .contains("headers string must decode to a JSON object or array"),
+            "unexpected error: {}",
+            err
+        );
+    }
+
+    #[test]
+    fn test_parse_timeout_secs_param_accepts_string_integer() {
+        let timeout = serde_json::json!("30");
+        assert_eq!(parse_timeout_secs_param(Some(&timeout)).unwrap(), Some(30));
+    }
+
+    #[test]
+    fn test_parse_timeout_secs_param_treats_empty_string_as_none() {
+        let timeout = serde_json::json!("");
+        assert_eq!(parse_timeout_secs_param(Some(&timeout)).unwrap(), None);
+    }
+
+    #[test]
+    fn test_parse_timeout_secs_param_rejects_value_above_cap() {
+        let timeout = serde_json::json!(MAX_TIMEOUT_SECS + 1);
+        let err = parse_timeout_secs_param(Some(&timeout)).unwrap_err();
+        assert!(
+            err.to_string()
+                .contains(&format!("timeout_secs must be <= {}", MAX_TIMEOUT_SECS)),
+            "unexpected error: {}",
+            err
+        );
+    }
+
+    #[test]
+    fn test_parse_timeout_secs_param_rejects_string_value_above_cap() {
+        let timeout = serde_json::json!((MAX_TIMEOUT_SECS + 1).to_string());
+        let err = parse_timeout_secs_param(Some(&timeout)).unwrap_err();
+        assert!(
+            err.to_string()
+                .contains(&format!("timeout_secs must be <= {}", MAX_TIMEOUT_SECS)),
+            "unexpected error: {}",
+            err
+        );
+    }
+
+    #[test]
+    fn test_parse_save_to_param_treats_empty_string_as_none() {
+        let save_to = serde_json::json!("");
+        assert_eq!(parse_save_to_param(Some(&save_to)).unwrap(), None);
+    }
+
    #[test]
    fn test_http_tool_schema_body_is_freeform() {
        let schema = HttpTool::new().parameters_schema();
@@ -907,12 +1067,22 @@ mod tests {
    // ── Approval requirement tests ──────────────────────────────────────

    #[test]
-    fn test_no_auth_headers_returns_unless_auto_approved() {
+    fn test_get_no_auth_headers_returns_never() {
        let tool = HttpTool::new();
        let params = serde_json::json!({
            "method": "GET",
            "url": "https://api.example.com/data"
        });
+        assert_eq!(tool.requires_approval(&params), ApprovalRequirement::Never);
+    }
+
+    #[test]
+    fn test_post_no_auth_headers_returns_unless_auto_approved() {
+        let tool = HttpTool::new();
+        let params = serde_json::json!({
+            "method": "POST",
+            "url": "https://api.example.com/data"
+        });
        assert_eq!(
            tool.requires_approval(&params),
            ApprovalRequirement::UnlessAutoApproved
@@ -996,21 +1166,18 @@ mod tests {
    }

    #[test]
-    fn test_non_auth_headers_return_unless_auto_approved() {
+    fn test_get_non_auth_headers_return_never() {
        let tool = HttpTool::new();
        let params = serde_json::json!({
            "method": "GET",
            "url": "https://example.com",
            "headers": {"Content-Type": "application/json", "Accept": "text/html"}
        });
-        assert_eq!(
-            tool.requires_approval(&params),
-            ApprovalRequirement::UnlessAutoApproved
-        );
+        assert_eq!(tool.requires_approval(&params), ApprovalRequirement::Never);
    }

    #[test]
-    fn test_empty_headers_return_unless_auto_approved() {
+    fn test_get_empty_headers_return_never() {
        let tool = HttpTool::new();

        // Empty object
@@ -1019,10 +1186,7 @@ mod tests {
            "url": "https://example.com",
            "headers": {}
        });
-        assert_eq!(
-            tool.requires_approval(&params),
-            ApprovalRequirement::UnlessAutoApproved
-        );
+        assert_eq!(tool.requires_approval(&params), ApprovalRequirement::Never);

        // Empty array
        let params = serde_json::json!({
@@ -1030,10 +1194,7 @@ mod tests {
            "url": "https://example.com",
            "headers": []
        });
-        assert_eq!(
-            tool.requires_approval(&params),
-            ApprovalRequirement::UnlessAutoApproved
-        );
+        assert_eq!(tool.requires_approval(&params), ApprovalRequirement::Never);
    }

    // ── Credential registry approval tests ─────────────────────────────
@@ -1063,7 +1224,7 @@ mod tests {
    }

    #[test]
-    fn test_host_without_credential_mapping_returns_unless_auto_approved() {
+    fn test_get_host_without_credential_mapping_returns_never() {
        use crate::tools::wasm::SharedCredentialRegistry;

        let registry = Arc::new(SharedCredentialRegistry::new());
@@ -1075,10 +1236,7 @@ mod tests {
            "method": "GET",
            "url": "https://api.example.com/data"
        });
-        assert_eq!(
-            tool.requires_approval(&params),
-            ApprovalRequirement::UnlessAutoApproved
-        );
+        assert_eq!(tool.requires_approval(&params), ApprovalRequirement::Never);
    }

    #[test]
@@ -1119,6 +1277,25 @@ mod tests {
        assert_eq!(extract_host_from_params(&params), None);
    }

+    #[test]
+    fn test_requires_approval_with_stringified_http_params() {
+        use crate::tools::wasm::SharedCredentialRegistry;
+
+        let tool = HttpTool::new().with_credentials(
+            Arc::new(SharedCredentialRegistry::new()),
+            Arc::new(test_secrets_store()),
+        );
+        let req = serde_json::json!({
+            "body": "",
+            "headers": "[]",
+            "method": "GET",
+            "save_to": "",
+            "timeout_secs": "30",
+            "url": "https://r.jina.ai/http://news.baidu.com/"
+        });
+        let _ = tool.requires_approval(&req);
+    }
+
    // ── DNS pinning tests ─────────────────────────────────────────────

    #[tokio::test]
@@ -8,7 +8,7 @@ use secrecy::{ExposeSecret, SecretString};

 use crate::context::JobContext;
 use crate::tools::builtin::path_utils::validate_path;
-use crate::tools::tool::{ApprovalRequirement, Tool, ToolError, ToolOutput};
+use crate::tools::tool::{Tool, ToolError, ToolOutput};

 /// Tool for analyzing images using a vision-capable model.
 pub struct ImageAnalyzeTool {
@@ -86,10 +86,6 @@ impl Tool for ImageAnalyzeTool {
        })
    }

-    fn requires_approval(&self, _params: &serde_json::Value) -> ApprovalRequirement {
-        ApprovalRequirement::UnlessAutoApproved
-    }
-
    fn requires_sanitization(&self) -> bool {
        true
    }
@@ -185,6 +181,7 @@ impl Tool for ImageAnalyzeTool {
 mod tests {
    use super::super::media_type_from_path;
    use super::*;
+    use crate::tools::tool::ApprovalRequirement;
    use tempfile::TempDir;

    #[test]
@@ -199,7 +196,7 @@ mod tests {
    }

    #[test]
-    fn test_requires_approval_returns_unless_auto_approved() {
+    fn test_requires_approval_returns_never() {
        let tool = ImageAnalyzeTool::new(
            "https://api.example.com".to_string(),
            "test-key".to_string(),
@@ -208,7 +205,7 @@ mod tests {
        );
        assert_eq!(
            tool.requires_approval(&serde_json::json!({})),
-            ApprovalRequirement::UnlessAutoApproved
+            ApprovalRequirement::Never
        );
    }

@@ -7,7 +7,7 @@ use secrecy::{ExposeSecret, SecretString};

 use crate::context::JobContext;
 use crate::tools::builtin::path_utils::validate_path;
-use crate::tools::tool::{ApprovalRequirement, Tool, ToolError, ToolOutput};
+use crate::tools::tool::{Tool, ToolError, ToolOutput};

 /// Tool for editing images using an AI image editing API.
 pub struct ImageEditTool {
@@ -85,10 +85,6 @@ impl Tool for ImageEditTool {
        })
    }

-    fn requires_approval(&self, _params: &serde_json::Value) -> ApprovalRequirement {
-        ApprovalRequirement::UnlessAutoApproved
-    }
-
    fn requires_sanitization(&self) -> bool {
        false
    }
@@ -266,6 +262,7 @@ impl ImageEditTool {
 #[cfg(test)]
 mod tests {
    use super::*;
+    use crate::tools::tool::ApprovalRequirement;
    use tempfile::TempDir;

    #[test]
@@ -280,7 +277,7 @@ mod tests {
        assert!(!tool.requires_sanitization());
        assert_eq!(
            tool.requires_approval(&serde_json::json!({})),
-            ApprovalRequirement::UnlessAutoApproved
+            ApprovalRequirement::Never
        );
    }

@@ -5,7 +5,6 @@ use secrecy::{ExposeSecret, SecretString};
 use serde::{Deserialize, Serialize};

 use crate::context::JobContext;
-use crate::tools::tool::ApprovalRequirement;
 use crate::tools::{Tool, ToolError, ToolOutput};

 /// Tool for generating images using FLUX or compatible image generation APIs.
@@ -87,10 +86,6 @@ impl Tool for ImageGenerateTool {
        })
    }

-    fn requires_approval(&self, _params: &serde_json::Value) -> ApprovalRequirement {
-        ApprovalRequirement::UnlessAutoApproved
-    }
-
    fn requires_sanitization(&self) -> bool {
        false
    }
@@ -186,6 +181,7 @@ impl Tool for ImageGenerateTool {
 #[cfg(test)]
 mod tests {
    use super::*;
+    use crate::tools::tool::ApprovalRequirement;

    #[test]
    fn test_tool_metadata() {
@@ -197,7 +193,7 @@ mod tests {
        assert_eq!(tool.name(), "image_generate");
        assert_eq!(
            tool.requires_approval(&serde_json::json!({})),
-            ApprovalRequirement::UnlessAutoApproved
+            ApprovalRequirement::Never
        );

        let schema = tool.parameters_schema();
@@ -12,6 +12,7 @@
 //! Use `memory_write` to persist important facts that should be remembered
 //! across sessions.

+use std::path::Path;
 use std::sync::Arc;

 use async_trait::async_trait;
@@ -26,6 +27,28 @@ use crate::workspace::{Workspace, paths};
 const PROTECTED_IDENTITY_FILES: &[&str] =
    &[paths::IDENTITY, paths::SOUL, paths::AGENTS, paths::USER];

+/// Detect paths that are clearly local filesystem references, not workspace-memory docs.
+///
+/// Examples:
+/// - `/Users/.../file.md` (Unix absolute)
+/// - `C:\Users\...` or `D:/work/...` (Windows absolute)
+/// - `~/notes.md` (home expansion shorthand)
+fn looks_like_filesystem_path(path: &str) -> bool {
+    if path.is_empty() {
+        return false;
+    }
+
+    if Path::new(path).is_absolute() || path.starts_with("~/") {
+        return true;
+    }
+
+    let bytes = path.as_bytes();
+    bytes.len() >= 3
+        && bytes[0].is_ascii_alphabetic()
+        && bytes[1] == b':'
+        && (bytes[2] == b'\\' || bytes[2] == b'/')
+}
+
 /// Tool for searching workspace memory.
 ///
 /// Performs hybrid search (FTS + semantic) across all memory documents.
@@ -143,7 +166,8 @@ impl Tool for MemoryWriteTool {
         be remembered across sessions. Targets: 'memory' for curated long-term facts, \
         'daily_log' for timestamped session notes, 'heartbeat' for the periodic \
         checklist (HEARTBEAT.md), 'bootstrap' to clear the first-run ritual file, \
-         or provide a custom path for arbitrary file creation."
+         or provide a custom workspace path for arbitrary file creation. \
+         Never pass absolute filesystem paths like '/Users/...' or 'C:\\...'."
    }

    fn parameters_schema(&self) -> serde_json::Value {
@@ -183,6 +207,14 @@ impl Tool for MemoryWriteTool {
            .and_then(|v| v.as_str())
            .unwrap_or("daily_log");

+        if looks_like_filesystem_path(target) {
+            return Err(ToolError::InvalidParameters(format!(
+                "'{}' looks like a local filesystem path. memory_write only works with workspace-memory paths. \
+                 Use write_file for filesystem writes. For opening files in an editor, use shell with: open \"<absolute_path>\".",
+                target
+            )));
+        }
+
        // Bootstrap target: clear BOOTSTRAP.md to mark first-run ritual complete.
        // Handled early because it accepts empty content (unlike other targets).
        if target == "bootstrap" {
@@ -332,7 +364,8 @@ impl Tool for MemoryReadTool {
    fn description(&self) -> &str {
        "Read a file from the workspace memory (database-backed storage). \
         Use this to read files shown by memory_tree. NOT for local filesystem files \
-         (use read_file for those). Works with identity files, heartbeat checklist, \
+         (use read_file for those). Do not pass absolute paths like '/Users/...' or 'C:\\...'. \
+         Works with identity files, heartbeat checklist, \
         memory, daily logs, or any custom workspace path."
    }

@@ -358,6 +391,14 @@ impl Tool for MemoryReadTool {

        let path = require_str(&params, "path")?;

+        if looks_like_filesystem_path(path) {
+            return Err(ToolError::InvalidParameters(format!(
+                "'{}' looks like a local filesystem path. memory_read only works with workspace-memory paths. \
+                 Use read_file for filesystem reads. For opening files in an editor, use shell with: open \"<absolute_path>\".",
+                path
+            )));
+        }
+
        let doc = self
            .workspace
            .read(path)
@@ -498,80 +539,100 @@ impl Tool for MemoryTreeTool {
    }
 }

-#[cfg(all(test, feature = "postgres"))]
+#[cfg(test)]
 mod tests {
    use super::*;

-    fn make_test_workspace() -> Arc<Workspace> {
-        Arc::new(Workspace::new(
-            "test_user",
-            deadpool_postgres::Pool::builder(deadpool_postgres::Manager::new(
-                tokio_postgres::Config::new(),
-                tokio_postgres::NoTls,
+    #[test]
+    fn detects_filesystem_paths() {
+        assert!(looks_like_filesystem_path("/Users/nige/file.md"));
+        assert!(looks_like_filesystem_path("C:\\Users\\nige\\file.md"));
+        assert!(looks_like_filesystem_path("D:/work/file.md"));
+        assert!(looks_like_filesystem_path("~/notes.md"));
+    }
+
+    #[test]
+    fn allows_workspace_memory_paths() {
+        assert!(!looks_like_filesystem_path("MEMORY.md"));
+        assert!(!looks_like_filesystem_path("daily/2026-03-11.md"));
+        assert!(!looks_like_filesystem_path("projects/alpha/notes.md"));
+    }
+
+    #[cfg(feature = "postgres")]
+    mod postgres_schema_tests {
+        use super::*;
+
+        fn make_test_workspace() -> Arc<Workspace> {
+            Arc::new(Workspace::new(
+                "test_user",
+                deadpool_postgres::Pool::builder(deadpool_postgres::Manager::new(
+                    tokio_postgres::Config::new(),
+                    tokio_postgres::NoTls,
+                ))
+                .build()
+                .unwrap(),
            ))
-            .build()
-            .unwrap(),
-        ))
-    }
+        }

-    #[test]
-    fn test_memory_search_schema() {
-        let workspace = make_test_workspace();
-        let tool = MemorySearchTool::new(workspace);
+        #[test]
+        fn test_memory_search_schema() {
+            let workspace = make_test_workspace();
+            let tool = MemorySearchTool::new(workspace);

-        assert_eq!(tool.name(), "memory_search");
-        assert!(!tool.requires_sanitization());
+            assert_eq!(tool.name(), "memory_search");
+            assert!(!tool.requires_sanitization());

-        let schema = tool.parameters_schema();
-        assert!(schema["properties"]["query"].is_object());
-        assert!(
-            schema["required"]
-                .as_array()
-                .unwrap()
-                .contains(&"query".into())
-        );
-    }
+            let schema = tool.parameters_schema();
+            assert!(schema["properties"]["query"].is_object());
+            assert!(
+                schema["required"]
+                    .as_array()
+                    .unwrap()
+                    .contains(&"query".into())
+            );
+        }

-    #[test]
-    fn test_memory_write_schema() {
-        let workspace = make_test_workspace();
-        let tool = MemoryWriteTool::new(workspace);
+        #[test]
+        fn test_memory_write_schema() {
+            let workspace = make_test_workspace();
+            let tool = MemoryWriteTool::new(workspace);

-        assert_eq!(tool.name(), "memory_write");
+            assert_eq!(tool.name(), "memory_write");

-        let schema = tool.parameters_schema();
-        assert!(schema["properties"]["content"].is_object());
-        assert!(schema["properties"]["target"].is_object());
-        assert!(schema["properties"]["append"].is_object());
-    }
+            let schema = tool.parameters_schema();
+            assert!(schema["properties"]["content"].is_object());
+            assert!(schema["properties"]["target"].is_object());
+            assert!(schema["properties"]["append"].is_object());
+        }

-    #[test]
-    fn test_memory_read_schema() {
-        let workspace = make_test_workspace();
-        let tool = MemoryReadTool::new(workspace);
+        #[test]
+        fn test_memory_read_schema() {
+            let workspace = make_test_workspace();
+            let tool = MemoryReadTool::new(workspace);

-        assert_eq!(tool.name(), "memory_read");
+            assert_eq!(tool.name(), "memory_read");

-        let schema = tool.parameters_schema();
-        assert!(schema["properties"]["path"].is_object());
-        assert!(
-            schema["required"]
-                .as_array()
-                .unwrap()
-                .contains(&"path".into())
-        );
-    }
+            let schema = tool.parameters_schema();
+            assert!(schema["properties"]["path"].is_object());
+            assert!(
+                schema["required"]
+                    .as_array()
+                    .unwrap()
+                    .contains(&"path".into())
+            );
+        }

-    #[test]
-    fn test_memory_tree_schema() {
-        let workspace = make_test_workspace();
-        let tool = MemoryTreeTool::new(workspace);
+        #[test]
+        fn test_memory_tree_schema() {
+            let workspace = make_test_workspace();
+            let tool = MemoryTreeTool::new(workspace);

-        assert_eq!(tool.name(), "memory_tree");
+            assert_eq!(tool.name(), "memory_tree");

-        let schema = tool.parameters_schema();
-        assert!(schema["properties"]["path"].is_object());
-        assert!(schema["properties"]["depth"].is_object());
-        assert_eq!(schema["properties"]["depth"]["default"], 1);
+            let schema = tool.parameters_schema();
+            assert!(schema["properties"]["path"].is_object());
+            assert!(schema["properties"]["depth"].is_object());
+            assert_eq!(schema["properties"]["depth"]["default"], 1);
+        }
    }
 }
@@ -15,6 +15,7 @@ pub mod secrets_tools;
 pub(crate) mod shell;
 pub mod skill_tools;
 mod time;
+mod tool_info;

 pub use echo::EchoTool;
 pub use extension_tools::{
@@ -39,6 +40,7 @@ pub use secrets_tools::{SecretDeleteTool, SecretListTool};
 pub use shell::ShellTool;
 pub use skill_tools::{SkillInstallTool, SkillListTool, SkillRemoveTool, SkillSearchTool};
 pub use time::TimeTool;
+pub use tool_info::ToolInfoTool;
 mod html_converter;
 pub mod image_analyze;
 pub mod image_edit;
@@ -104,6 +104,14 @@ impl Tool for RoutineCreateTool {
                    "enum": ["lightweight", "full_job"],
                    "description": "Execution mode: 'lightweight' (single LLM call, default) or 'full_job' (multi-turn with tools)"
                },
+                "use_tools": {
+                    "type": "boolean",
+                    "description": "Enable tool access in lightweight mode (default: false). Only safe tools (no approval required) are available. Ignored for full_job mode."
+                },
+                "max_tool_rounds": {
+                    "type": "integer",
+                    "description": "Max tool call rounds in lightweight mode (default: 3). Only used when use_tools is true."
+                },
                "cooldown_secs": {
                    "type": "integer",
                    "description": "Minimum seconds between fires (default: 300)"
@@ -262,11 +270,24 @@ impl Tool for RoutineCreateTool {
            })
            .unwrap_or_default();

+        let use_tools = params
+            .get("use_tools")
+            .and_then(|v| v.as_bool())
+            .unwrap_or(false);
+
+        let max_tool_rounds = params
+            .get("max_tool_rounds")
+            .and_then(|v| v.as_u64())
+            .map(|v| v.clamp(1, crate::agent::routine::MAX_TOOL_ROUNDS_LIMIT as u64) as u32)
+            .unwrap_or(3);
+
        let action = match action_type {
            "lightweight" => RoutineAction::Lightweight {
                prompt: prompt.to_string(),
                context_paths,
                max_tokens: 4096,
+                use_tools,
+                max_tool_rounds,
            },
            "full_job" => {
                let tool_permissions = crate::agent::routine::parse_tool_permissions(&params);
@@ -0,0 +1,183 @@
+//! On-demand tool discovery (like CLI `--help`).
+//!
+//! Two levels of detail:
+//! - Default: name, description, parameter names (compact ~150 bytes)
+//! - `include_schema: true`: adds the full typed JSON Schema
+//!
+//! Keeps the tools array compact (WASM tools use permissive schemas)
+//! while allowing precise discovery when needed.
+
+use std::sync::Weak;
+
+use async_trait::async_trait;
+
+use crate::context::JobContext;
+use crate::tools::registry::ToolRegistry;
+use crate::tools::tool::{Tool, ToolError, ToolOutput, require_str};
+
+pub struct ToolInfoTool {
+    registry: Weak<ToolRegistry>,
+}
+
+impl ToolInfoTool {
+    pub fn new(registry: Weak<ToolRegistry>) -> Self {
+        Self { registry }
+    }
+}
+
+#[async_trait]
+impl Tool for ToolInfoTool {
+    fn name(&self) -> &str {
+        "tool_info"
+    }
+
+    fn description(&self) -> &str {
+        "Get info about any tool: description and parameter names. \
+         Set include_schema to true for the full typed parameter schema."
+    }
+
+    fn parameters_schema(&self) -> serde_json::Value {
+        serde_json::json!({
+            "type": "object",
+            "properties": {
+                "name": {
+                    "type": "string",
+                    "description": "Name of the tool to get info about"
+                },
+                "include_schema": {
+                    "type": "boolean",
+                    "description": "If true, include the full typed JSON Schema for parameters (larger response). Default: false.",
+                    "default": false
+                }
+            },
+            "required": ["name"]
+        })
+    }
+
+    async fn execute(
+        &self,
+        params: serde_json::Value,
+        _ctx: &JobContext,
+    ) -> Result<ToolOutput, ToolError> {
+        let start = std::time::Instant::now();
+        let name = require_str(&params, "name")?;
+        let include_schema = params
+            .get("include_schema")
+            .and_then(|v| v.as_bool())
+            .unwrap_or(false);
+
+        let registry = self.registry.upgrade().ok_or_else(|| {
+            ToolError::ExecutionFailed(
+                "tool registry is no longer available for tool_info".to_string(),
+            )
+        })?;
+
+        let tool = registry.get(name).await.ok_or_else(|| {
+            ToolError::InvalidParameters(format!("No tool named '{name}' is registered"))
+        })?;
+
+        let schema = tool.discovery_schema();
+
+        // Extract just param names from the schema's "properties" keys
+        let param_names: Vec<&str> = schema
+            .get("properties")
+            .and_then(|p| p.as_object())
+            .map(|props| props.keys().map(|k| k.as_str()).collect())
+            .unwrap_or_default();
+
+        let mut info = serde_json::json!({
+            "name": tool.name(),
+            "description": tool.description(),
+            "parameters": param_names,
+        });
+
+        if include_schema {
+            info["schema"] = schema;
+        }
+
+        Ok(ToolOutput::success(info, start.elapsed()))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::tools::builtin::EchoTool;
+    use std::sync::Arc;
+
+    #[tokio::test]
+    async fn test_tool_info_default_returns_param_names() {
+        let registry = Arc::new(ToolRegistry::new());
+        registry.register(Arc::new(EchoTool)).await;
+
+        let tool = ToolInfoTool::new(Arc::downgrade(&registry));
+        let ctx = JobContext::default();
+        let result = tool
+            .execute(serde_json::json!({"name": "echo"}), &ctx)
+            .await
+            .unwrap();
+
+        let info = &result.result;
+        assert_eq!(info["name"], "echo");
+        assert!(!info["description"].as_str().unwrap().is_empty());
+        // Default: parameters is an array of names, not the full schema
+        assert!(info["parameters"].is_array());
+        assert!(
+            info["parameters"]
+                .as_array()
+                .unwrap()
+                .iter()
+                .any(|v| v.as_str() == Some("message")),
+            "echo tool should have 'message' parameter: {:?}",
+            info["parameters"]
+        );
+        // No schema field by default
+        assert!(info.get("schema").is_none());
+    }
+
+    #[tokio::test]
+    async fn test_tool_info_with_schema() {
+        let registry = Arc::new(ToolRegistry::new());
+        registry.register(Arc::new(EchoTool)).await;
+
+        let tool = ToolInfoTool::new(Arc::downgrade(&registry));
+        let ctx = JobContext::default();
+        let result = tool
+            .execute(
+                serde_json::json!({"name": "echo", "include_schema": true}),
+                &ctx,
+            )
+            .await
+            .unwrap();
+
+        let info = &result.result;
+        assert_eq!(info["name"], "echo");
+        // With include_schema: true, schema field should be present
+        assert!(info["schema"].is_object());
+        assert!(info["schema"]["properties"].is_object());
+    }
+
+    #[tokio::test]
+    async fn test_tool_info_unknown_tool() {
+        let registry = Arc::new(ToolRegistry::new());
+        let tool = ToolInfoTool::new(Arc::downgrade(&registry));
+        let ctx = JobContext::default();
+        let result = tool
+            .execute(serde_json::json!({"name": "nonexistent"}), &ctx)
+            .await;
+        assert!(result.is_err());
+    }
+
+    #[tokio::test]
+    async fn test_tool_info_registry_dropped() {
+        let registry = Arc::new(ToolRegistry::new());
+        let tool = ToolInfoTool::new(Arc::downgrade(&registry));
+        drop(registry);
+
+        let ctx = JobContext::default();
+        let result = tool
+            .execute(serde_json::json!({"name": "echo"}), &ctx)
+            .await;
+        assert!(matches!(result, Err(ToolError::ExecutionFailed(_))));
+    }
+}
@@ -669,7 +669,7 @@ pub async fn authorize_mcp_server(
    }

    // Determine client_id and endpoints
-    let (client_id, authorization_url, token_url, use_pkce, scopes, extra_params) =
+    let (client_id, authorization_url, token_url, use_pkce, scopes, mut extra_params) =
        if let Some(oauth) = &server_config.oauth {
            // Pre-configured OAuth
            let (auth_url, tok_url) = discover_oauth_endpoints(server_config).await?;
@@ -711,6 +711,13 @@ pub async fn authorize_mcp_server(
        None
    };

+    // Generate OAuth state parameter. While optional in OAuth 2.1 with PKCE,
+    // some MCP servers (e.g. Attio) require it.
+    let mut state_bytes = [0u8; 16];
+    rand::rngs::OsRng.fill_bytes(&mut state_bytes);
+    let state = URL_SAFE_NO_PAD.encode(state_bytes);
+    extra_params.insert("state".to_string(), state);
+
    // Compute canonical resource URI for RFC 8707
    let resource = canonical_resource_uri(&server_config.url);

@@ -741,7 +748,10 @@ pub async fn authorize_mcp_server(

    println!("  Waiting for authorization...");

-    // Wait for callback
+    // Wait for callback. State is sent in the URL for servers that require it
+    // (e.g. Attio), but we don't enforce validation on the callback because MCP
+    // servers use PKCE which already binds the request to the token exchange,
+    // and some servers may not echo state back.
    let code = wait_for_authorization_callback(listener, &server_config.name).await?;

    println!("  Exchanging code for token...");
@@ -1711,4 +1721,69 @@ mod tests {

        assert!(!url.contains("resource="));
    }
+
+    /// Regression test: MCP OAuth authorization URLs must include a `state`
+    /// parameter. While OAuth 2.1 makes `state` optional when PKCE is used,
+    /// some MCP servers (e.g. Attio) require it and reject requests without it:
+    /// {"error":"invalid_request","error_description":"Invalid value provided
+    /// for: state"}
+    ///
+    /// Including `state` is harmless for servers that don't require it, since
+    /// it is a standard OAuth parameter that compliant servers will echo back
+    /// or ignore.
+    ///
+    /// The state is generated in `authorize_mcp_server` and injected into
+    /// `extra_params` before `build_authorization_url` is called. This test
+    /// verifies that `build_authorization_url` correctly propagates state from
+    /// extra_params into the URL, and that each generated state is unique.
+    #[test]
+    fn test_authorization_url_includes_state_parameter() {
+        // Simulate what authorize_mcp_server does: generate state and
+        // insert it into extra_params.
+        let mut extra_params = HashMap::new();
+        let mut state_bytes = [0u8; 16];
+        rand::rngs::OsRng.fill_bytes(&mut state_bytes);
+        let state = URL_SAFE_NO_PAD.encode(state_bytes);
+        extra_params.insert("state".to_string(), state.clone());
+
+        let pkce = PkceChallenge::generate();
+        let url = build_authorization_url(
+            "https://app.attio.com/oidc/authorize",
+            "test-client",
+            "http://127.0.0.1:9876/callback",
+            &[
+                "mcp".to_string(),
+                "offline_access".to_string(),
+                "openid".to_string(),
+            ],
+            Some(&pkce),
+            &extra_params,
+            Some("https://mcp.attio.com/mcp"),
+        );
+
+        // State must be present in the URL
+        assert!(
+            url.contains(&format!("state={}", state)),
+            "Authorization URL must include the state parameter, got: {}",
+            url,
+        );
+
+        // State must be base64url-encoded (no padding, no +/)
+        assert!(!state.contains('+'), "State must be base64url-safe");
+        assert!(!state.contains('/'), "State must be base64url-safe");
+        assert!(!state.contains('='), "State must not have padding");
+
+        // State must have sufficient entropy (16 bytes -> 22 base64url chars)
+        assert!(
+            state.len() >= 22,
+            "State must have at least 128 bits of entropy, got {} chars",
+            state.len(),
+        );
+
+        // Two generated states must differ
+        let mut state_bytes_2 = [0u8; 16];
+        rand::rngs::OsRng.fill_bytes(&mut state_bytes_2);
+        let state_2 = URL_SAFE_NO_PAD.encode(state_bytes_2);
+        assert_ne!(state, state_2, "State must be unique per request");
+    }
 }
@@ -12,6 +12,7 @@ pub mod builtin;
 pub mod execute;
 pub mod mcp;
 pub mod rate_limiter;
+pub mod redaction;
 pub mod schema_validator;
 pub mod wasm;

@@ -0,0 +1,251 @@
+use serde_json::{Map, Value};
+
+const REDACTED: &str = "[REDACTED]";
+const SENSITIVE_EXACT: &[&str] = &[
+    "authorization",
+    "proxy-authorization",
+    "cookie",
+    "set-cookie",
+    "x-api-key",
+    "api-key",
+    "api_key",
+    "access_token",
+    "refresh_token",
+    "session_token",
+    "id_token",
+    "token",
+    "password",
+    "passwd",
+    "secret",
+    "client_secret",
+    "private_key",
+    "apikey",
+    "apisecret",
+];
+
+const SENSITIVE_PARTS: &[&str] = &[
+    "password",
+    "passwd",
+    "secret",
+    "credential",
+    "authorization",
+    "cookie",
+    "apikey",
+    "apisecret",
+];
+const TOKEN_PARTS: &[&str] = &["token", "jwt"];
+const KEY_PARTS: &[&str] = &["key"];
+const CONTEXT_PARTS: &[&str] = &[
+    "auth",
+    "oauth",
+    "authorization",
+    "api",
+    "access",
+    "refresh",
+    "session",
+    "bearer",
+    "private",
+    "client",
+    "id",
+    "app",
+    "user",
+    "application",
+    "account",
+];
+
+fn split_camel_case_key_parts(key: &str) -> Vec<String> {
+    if key.is_empty() {
+        return Vec::new();
+    }
+
+    let chars: Vec<char> = key.chars().collect();
+    let mut parts = Vec::new();
+    let mut start = 0;
+
+    for i in 1..chars.len() {
+        let prev = chars[i - 1];
+        let cur = chars[i];
+        let next = chars.get(i + 1).copied();
+
+        let boundary = (prev.is_ascii_lowercase() && cur.is_ascii_uppercase())
+            || (prev.is_ascii_alphabetic() && cur.is_ascii_digit())
+            || (prev.is_ascii_digit() && cur.is_ascii_alphabetic())
+            || (prev.is_ascii_uppercase()
+                && cur.is_ascii_uppercase()
+                && next.map(|n| n.is_ascii_lowercase()).unwrap_or(false));
+
+        if boundary {
+            parts.push(chars[start..i].iter().collect::<String>());
+            start = i;
+        }
+    }
+
+    parts.push(chars[start..].iter().collect::<String>());
+    parts
+}
+
+fn tokenize_key_parts(key: &str) -> Vec<String> {
+    let mut parts = Vec::new();
+
+    for segment in key.split(|c: char| !c.is_ascii_alphanumeric()) {
+        if segment.is_empty() {
+            continue;
+        }
+
+        parts.extend(split_camel_case_key_parts(segment));
+    }
+
+    parts.into_iter().map(|p| p.to_ascii_lowercase()).collect()
+}
+
+fn has_exact(parts: &[String], candidates: &[&str]) -> bool {
+    parts
+        .iter()
+        .any(|part| candidates.iter().any(|candidate| part == candidate))
+}
+
+fn has_candidate_or_numbered_variant(parts: &[String], candidates: &[&str]) -> bool {
+    parts.iter().any(|part| {
+        candidates.iter().any(|candidate| {
+            if part == candidate {
+                return true;
+            }
+            let Some(suffix) = part.strip_prefix(candidate) else {
+                return false;
+            };
+            !suffix.is_empty() && suffix.chars().all(|c| c.is_ascii_digit())
+        })
+    })
+}
+
+fn has_contextual_suffix(parts: &[String], candidates: &[&str]) -> bool {
+    parts.iter().any(|part| {
+        candidates.iter().any(|candidate| {
+            let Some(prefix) = part.strip_suffix(candidate) else {
+                return false;
+            };
+            !prefix.is_empty() && CONTEXT_PARTS.contains(&prefix)
+        })
+    })
+}
+
+fn is_sensitive_key(key: &str) -> bool {
+    let lower = key.to_ascii_lowercase();
+    if SENSITIVE_EXACT.contains(&lower.as_str()) {
+        return true;
+    }
+
+    let parts = tokenize_key_parts(key);
+    if parts.is_empty() {
+        return false;
+    }
+
+    if has_candidate_or_numbered_variant(&parts, SENSITIVE_PARTS) {
+        return true;
+    }
+
+    let has_token = has_candidate_or_numbered_variant(&parts, TOKEN_PARTS);
+    let has_key = has_candidate_or_numbered_variant(&parts, KEY_PARTS);
+
+    if has_token && has_key {
+        return true;
+    }
+
+    if has_contextual_suffix(&parts, TOKEN_PARTS) || has_contextual_suffix(&parts, KEY_PARTS) {
+        return true;
+    }
+
+    let has_context = has_exact(&parts, CONTEXT_PARTS);
+    has_context && (has_token || has_key)
+}
+
+fn redact_in_place(value: &mut Value) {
+    match value {
+        Value::Object(map) => redact_object(map),
+        Value::Array(items) => {
+            for item in items {
+                redact_in_place(item);
+            }
+        }
+        _ => {}
+    }
+}
+
+fn redact_object(map: &mut Map<String, Value>) {
+    for (key, val) in map {
+        if is_sensitive_key(key) {
+            *val = Value::String(REDACTED.to_string());
+        } else {
+            redact_in_place(val);
+        }
+    }
+}
+
+pub fn redact_sensitive_json(value: &Value) -> Value {
+    let mut cloned = value.clone();
+    redact_in_place(&mut cloned);
+    cloned
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{is_sensitive_key, redact_sensitive_json};
+
+    #[test]
+    fn redacts_exact_sensitive_keys() {
+        let input = serde_json::json!({
+            "headers": {
+                "Authorization": "Bearer abc",
+                "x-api-key": "k-123",
+                "content-type": "application/json"
+            },
+            "password": "p@ss"
+        });
+        let out = redact_sensitive_json(&input);
+        assert_eq!(out["headers"]["Authorization"], "[REDACTED]");
+        assert_eq!(out["headers"]["x-api-key"], "[REDACTED]");
+        assert_eq!(out["headers"]["content-type"], "application/json");
+        assert_eq!(out["password"], "[REDACTED]");
+    }
+
+    #[test]
+    fn redacts_nested_sensitive_keys() {
+        let input = serde_json::json!({
+            "body": {
+                "clientSecret": "xyz",
+                "nested": [{"authToken": "123"}, {"query": "ok"}]
+            }
+        });
+        let out = redact_sensitive_json(&input);
+        assert_eq!(out["body"]["clientSecret"], "[REDACTED]");
+        assert_eq!(out["body"]["nested"][0]["authToken"], "[REDACTED]");
+        assert_eq!(out["body"]["nested"][1]["query"], "ok");
+    }
+
+    #[test]
+    fn does_not_over_redact_common_non_sensitive_keys() {
+        assert!(!is_sensitive_key("author"));
+        assert!(!is_sensitive_key("authorize_user"));
+        assert!(!is_sensitive_key("token_count"));
+        assert!(!is_sensitive_key("tokenize"));
+        assert!(!is_sensitive_key("oauth_redirect_uri"));
+    }
+
+    #[test]
+    fn still_redacts_expected_token_keys() {
+        assert!(is_sensitive_key("auth_token"));
+        assert!(is_sensitive_key("oauth_token"));
+        assert!(is_sensitive_key("accessToken"));
+        assert!(is_sensitive_key("apiKey"));
+        assert!(is_sensitive_key("token_key"));
+        assert!(is_sensitive_key("appTokenKey"));
+        assert!(is_sensitive_key("userJwt"));
+    }
+
+    #[test]
+    fn redacts_lowercase_digit_suffix_segments() {
+        assert!(is_sensitive_key("password123"));
+        assert!(is_sensitive_key("secret99"));
+        assert!(is_sensitive_key("accounttoken2"));
+    }
+}
@@ -23,7 +23,7 @@ use crate::tools::builtin::{
    ToolUpgradeTool, WriteFileTool,
 };
 use crate::tools::rate_limiter::RateLimiter;
-use crate::tools::tool::{Tool, ToolDomain};
+use crate::tools::tool::{ApprovalRequirement, Tool, ToolDomain};
 use crate::tools::wasm::{
    Capabilities, OAuthRefreshConfig, ResourceLimits, SharedCredentialRegistry, WasmError,
    WasmStorageError, WasmToolRuntime, WasmToolStore, WasmToolWrapper,
@@ -75,6 +75,7 @@ const PROTECTED_TOOL_NAMES: &[&str] = &[
    "image_generate",
    "image_edit",
    "image_analyze",
+    "tool_info",
 ];

 /// Registry of available tools.
@@ -245,6 +246,17 @@ impl ToolRegistry {
        tracing::debug!("Registered {} built-in tools", self.count());
    }

+    /// Register the `tool_info` discovery tool.
+    ///
+    /// Requires `Arc<Self>` so the tool can query the registry for other tools'
+    /// schemas at runtime. Call after `register_builtin_tools()`.
+    pub fn register_tool_info(self: &Arc<Self>) {
+        use crate::tools::builtin::ToolInfoTool;
+        let tool = ToolInfoTool::new(Arc::downgrade(self));
+        self.register_sync(Arc::new(tool));
+        tracing::debug!("Registered tool_info discovery tool");
+    }
+
    /// Register only orchestrator-domain tools (safe for the main process).
    ///
    /// This registers tools that don't touch the filesystem or run shell commands:
@@ -278,6 +290,38 @@ impl ToolRegistry {
            .collect()
    }

+    /// Get tool definitions excluding specific tools by name.
+    ///
+    /// Used by lightweight routines to filter out denylisted and approval-gated tools
+    /// so the LLM only sees tools it is actually allowed to call.
+    pub async fn tool_definitions_excluding(&self, deny: &[&str]) -> Vec<ToolDefinition> {
+        let empty_params = serde_json::Value::Object(serde_json::Map::new());
+        let mut defs: Vec<ToolDefinition> = self
+            .tools
+            .read()
+            .await
+            .values()
+            .filter(|tool| {
+                // Exclude denylisted tools
+                if deny.contains(&tool.name()) {
+                    return false;
+                }
+                // Exclude tools that require approval
+                matches!(
+                    tool.requires_approval(&empty_params),
+                    ApprovalRequirement::Never
+                )
+            })
+            .map(|tool| ToolDefinition {
+                name: tool.name().to_string(),
+                description: tool.description().to_string(),
+                parameters: tool.parameters_schema(),
+            })
+            .collect();
+        defs.sort_unstable_by(|a, b| a.name.cmp(&b.name));
+        defs
+    }
+
    /// Register development tools for building software.
    ///
    /// These tools provide shell access, file operations, and code editing
@@ -336,6 +336,17 @@ pub trait Tool: Send + Sync {
        None
    }

+    /// Full parameter schema for discovery and coercion purposes.
+    ///
+    /// Unlike `parameters_schema()` (which may be permissive to keep the tools
+    /// array compact), this returns the complete typed schema. Used by the
+    /// `tool_info` built-in and by WASM parameter coercion.
+    ///
+    /// Default: delegates to `parameters_schema()`.
+    fn discovery_schema(&self) -> serde_json::Value {
+        self.parameters_schema()
+    }
+
    /// Get the tool schema for LLM function calling.
    fn schema(&self) -> ToolSchema {
        ToolSchema {
@@ -1,7 +1,5 @@
 //! WASM sandbox error types.

-use std::fmt;
-
 use thiserror::Error;

 /// Errors that can occur during WASM tool execution.
@@ -68,13 +66,13 @@ pub enum WasmError {
    Timeout(std::time::Duration),

    /// Component returned an error response.
-    /// When `hint` is non-empty it carries the tool's description and parameter
-    /// schema so the LLM can retry with correct arguments.
+    /// When `hint` is non-empty it points the LLM to `tool_info` so it can
+    /// fetch the tool's full parameter schema on demand.
    #[error("Tool error: {message}{}", if hint.is_empty() { String::new() } else { format!("\n\nTool usage hint:\n{hint}") })]
    ToolReturnedError {
        /// The error message from the WASM tool.
        message: String,
-        /// Optional description + schema hint (empty when unavailable).
+        /// Optional retry hint (empty when unavailable).
        hint: String,
    },

@@ -99,73 +97,9 @@ impl From<WasmError> for crate::tools::ToolError {
    }
 }

-/// Details about a trap that occurred during execution.
-#[derive(Debug, Clone)]
-pub struct TrapInfo {
-    /// Human-readable trap message.
-    pub message: String,
-    /// Trap code if available.
-    pub code: Option<TrapCode>,
-}
-
-impl fmt::Display for TrapInfo {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        match &self.code {
-            Some(code) => write!(f, "{}: {}", code, self.message),
-            None => write!(f, "{}", self.message),
-        }
-    }
-}
-
-/// Known trap codes from Wasmtime.
-#[derive(Debug, Clone, Copy, PartialEq, Eq)]
-pub enum TrapCode {
-    /// Out of bounds memory access.
-    MemoryOutOfBounds,
-    /// Out of bounds table access.
-    TableOutOfBounds,
-    /// Indirect call type mismatch.
-    IndirectCallToNull,
-    /// Signature mismatch on indirect call.
-    BadSignature,
-    /// Integer overflow.
-    IntegerOverflow,
-    /// Integer division by zero.
-    IntegerDivisionByZero,
-    /// Invalid conversion to integer.
-    BadConversionToInteger,
-    /// Unreachable instruction executed.
-    UnreachableCodeReached,
-    /// Call stack exhausted.
-    StackOverflow,
-    /// Out of fuel.
-    OutOfFuel,
-    /// Unknown trap code.
-    Unknown,
-}
-
-impl fmt::Display for TrapCode {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        let s = match self {
-            TrapCode::MemoryOutOfBounds => "memory out of bounds",
-            TrapCode::TableOutOfBounds => "table out of bounds",
-            TrapCode::IndirectCallToNull => "indirect call to null",
-            TrapCode::BadSignature => "bad signature",
-            TrapCode::IntegerOverflow => "integer overflow",
-            TrapCode::IntegerDivisionByZero => "integer division by zero",
-            TrapCode::BadConversionToInteger => "bad conversion to integer",
-            TrapCode::UnreachableCodeReached => "unreachable code reached",
-            TrapCode::StackOverflow => "stack overflow",
-            TrapCode::OutOfFuel => "out of fuel",
-            TrapCode::Unknown => "unknown trap",
-        };
-        write!(f, "{}", s)
-    }
-}
-
 #[cfg(test)]
 mod tests {
-    use crate::tools::wasm::error::{TrapCode, TrapInfo, WasmError};
+    use crate::tools::wasm::error::WasmError;

    #[test]
    fn test_error_display() {
@@ -180,17 +114,6 @@ mod tests {
        assert!(err.to_string().contains("10000000"));
    }

-    #[test]
-    fn test_trap_info_display() {
-        let info = TrapInfo {
-            message: "access at offset 0x1000".to_string(),
-            code: Some(TrapCode::MemoryOutOfBounds),
-        };
-        let s = info.to_string();
-        assert!(s.contains("memory out of bounds"));
-        assert!(s.contains("access at offset"));
-    }
-
    #[test]
    fn test_conversion_to_tool_error() {
        let wasm_err = WasmError::Trapped("test trap".to_string());
@@ -218,12 +141,11 @@ mod tests {
    fn test_tool_returned_error_with_hint() {
        let err = WasmError::ToolReturnedError {
            message: "unknown action: foobar".to_string(),
-            hint: "Description: Gmail tool\nParameters schema: {\"type\":\"object\"}".to_string(),
+            hint: "Tip: call tool_info(name: \"gmail\", include_schema: true) for the full parameter schema.".to_string(),
        };
        let display = err.to_string();
        assert!(display.contains("unknown action: foobar"));
        assert!(display.contains("Tool usage hint"));
-        assert!(display.contains("Gmail tool"));
-        assert!(display.contains("Parameters schema"));
+        assert!(display.contains("tool_info"));
    }
 }
@@ -67,14 +67,8 @@ pub struct WasmResourceLimiter {
    memory_used: u64,
    /// Maximum tables allowed.
    max_tables: u32,
-    /// Current table count.
-    #[allow(dead_code)] // Reserved for table limit enforcement
-    tables_created: u32,
    /// Maximum instances allowed.
    max_instances: u32,
-    /// Current instance count.
-    #[allow(dead_code)] // Reserved for instance limit enforcement
-    instances_created: u32,
 }

 impl WasmResourceLimiter {
@@ -87,9 +81,7 @@ impl WasmResourceLimiter {
            memory_limit,
            memory_used: 0,
            max_tables: 10,
-            tables_created: 0,
            max_instances: 10, // Component model needs multiple instances for WASI
-            instances_created: 0,
        }
    }

@@ -96,7 +96,7 @@ pub(crate) mod storage;
 mod wrapper;

 // Core types
-pub use error::{TrapCode, TrapInfo, WasmError};
+pub use error::WasmError;
 pub use host::{HostState, LogEntry, LogLevel};
 pub use limits::{
    DEFAULT_FUEL_LIMIT, DEFAULT_MEMORY_LIMIT, DEFAULT_TIMEOUT, FuelConfig, ResourceLimits,
@@ -123,7 +123,9 @@ pub struct PreparedModule {
    pub name: String,
    /// Tool description (cached from component).
    pub description: String,
-    /// Parameter schema JSON (cached from component).
+    /// Full parameter schema JSON extracted from the component.
+    /// Used for discovery and coercion, not necessarily for the compact
+    /// schema advertised in the main tools array.
    pub schema: serde_json::Value,
    /// Pre-compiled component (cheaply cloneable via internal Arc).
    component: wasmtime::component::Component,
@@ -265,11 +267,29 @@ impl WasmToolRuntime {
            let component = wasmtime::component::Component::new(&engine, &wasm_bytes)
                .map_err(|e| WasmError::CompilationFailed(e.to_string()))?;

-            // We need to instantiate briefly to extract metadata.
-            // In a full implementation, we'd use WIT bindgen to get typed access.
-            // For now, we extract what we can from the component.
-            let description = extract_tool_description(&engine, &component)?;
-            let schema = extract_tool_schema(&engine, &component)?;
+            // Briefly instantiate to extract metadata (description + schema)
+            // from the tool's exports, analogous to MCP's list_tools().
+            let effective_limits = limits.clone().unwrap_or(default_limits.clone());
+            let (description, schema) = crate::tools::wasm::wrapper::extract_wasm_metadata(
+                &engine,
+                &component,
+                &effective_limits,
+            )
+            .unwrap_or_else(|e| {
+                tracing::warn!(
+                    name = %name,
+                    error = %e,
+                    "WASM metadata extraction failed, using fallbacks"
+                );
+                (
+                    "WASM sandboxed tool".to_string(),
+                    serde_json::json!({
+                        "type": "object",
+                        "properties": {},
+                        "additionalProperties": true
+                    }),
+                )
+            });

            Ok::<_, WasmError>(PreparedModule {
                name: name.clone(),
@@ -321,41 +341,6 @@ impl WasmToolRuntime {
    }
 }

-/// Extract tool description from a compiled component.
-///
-/// Returns a generic fallback. Callers should prefer loading the description
-/// from the sidecar `*.capabilities.json` file and overriding via
-/// `WasmToolWrapper::with_description()` or the `WasmToolRegistration::description` field.
-fn extract_tool_description(
-    _engine: &Engine,
-    _component: &wasmtime::component::Component,
-) -> Result<String, WasmError> {
-    // WIT bindgen extraction is not yet implemented (see TODO #4 in CLAUDE.md).
-    // Real descriptions come from the capabilities.json sidecar file, which is
-    // loaded by the WasmToolLoader and passed as an override at registration time.
-    Ok("WASM sandboxed tool".to_string())
-}
-
-/// Extract tool parameter schema from a compiled component.
-///
-/// Returns a permissive fallback that accepts any JSON object. Callers should
-/// prefer loading the schema from the sidecar `*.capabilities.json` file and
-/// overriding via `WasmToolWrapper::with_schema()` or the
-/// `WasmToolRegistration::schema` field.
-fn extract_tool_schema(
-    _engine: &Engine,
-    _component: &wasmtime::component::Component,
-) -> Result<serde_json::Value, WasmError> {
-    // WIT bindgen extraction is not yet implemented (see TODO #4 in CLAUDE.md).
-    // Real schemas come from the capabilities.json sidecar file, which is
-    // loaded by the WasmToolLoader and passed as an override at registration time.
-    Ok(serde_json::json!({
-        "type": "object",
-        "properties": {},
-        "additionalProperties": true
-    }))
-}
-
 impl std::fmt::Debug for WasmToolRuntime {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        f.debug_struct("WasmToolRuntime")
@@ -464,9 +464,10 @@ pub struct WasmToolWrapper {
    /// Capabilities to grant to this tool.
    capabilities: Capabilities,
    /// Cached description (from PreparedModule or override).
+    /// Stored without any tool_info hints — hints are composed at display time.
    description: String,
-    /// Cached schema (from PreparedModule or override).
-    schema: serde_json::Value,
+    /// Compact and discovery schemas for this tool.
+    schemas: WasmToolSchemas,
    /// Injected credentials for HTTP requests (e.g., OAuth tokens).
    /// Keys are placeholder names like "GOOGLE_ACCESS_TOKEN".
    credentials: HashMap<String, String>,
@@ -477,6 +478,84 @@ pub struct WasmToolWrapper {
    oauth_refresh: Option<OAuthRefreshConfig>,
 }

+#[derive(Debug, Clone)]
+struct WasmToolSchemas {
+    /// Compact schema advertised in the main tools array.
+    ///
+    /// This stays permissive by default to avoid serializing full exported
+    /// WASM schemas on every LLM call. Sidecars can override it explicitly.
+    advertised: serde_json::Value,
+    /// Full schema available for discovery and coercion.
+    ///
+    /// Seeded from the WASM `schema()` export at registration time, unless a
+    /// sidecar explicitly overrides it.
+    discovery: serde_json::Value,
+}
+
+impl WasmToolSchemas {
+    fn permissive_schema() -> serde_json::Value {
+        serde_json::json!({
+            "type": "object",
+            "properties": {},
+            "additionalProperties": true
+        })
+    }
+
+    fn is_permissive_schema(schema: &serde_json::Value) -> bool {
+        schema
+            .get("properties")
+            .and_then(|p| p.as_object())
+            .is_none_or(|p| p.is_empty())
+    }
+
+    fn new(discovery: serde_json::Value) -> Self {
+        Self {
+            advertised: Self::permissive_schema(),
+            discovery,
+        }
+    }
+
+    fn with_override(&self, schema: serde_json::Value) -> Self {
+        Self {
+            advertised: schema.clone(),
+            discovery: schema,
+        }
+    }
+
+    fn is_advertised_permissive(&self) -> bool {
+        Self::is_permissive_schema(&self.advertised)
+    }
+
+    fn advertised(&self) -> serde_json::Value {
+        self.advertised.clone()
+    }
+
+    fn discovery(&self) -> serde_json::Value {
+        self.discovery.clone()
+    }
+
+    /// Return the best schema available for type coercion.
+    ///
+    /// Prefers the discovery schema when it has typed properties. Falls back
+    /// to the `PreparedModule` schema extracted at load time rather than
+    /// re-calling the WASM `schema()` export mid-execution, which could
+    /// interact with mutable linear memory state.
+    fn effective_for_coercion(&self, prepared_schema: &serde_json::Value) -> serde_json::Value {
+        if !Self::is_permissive_schema(&self.discovery) {
+            return self.discovery.clone();
+        }
+
+        // Fall back to the load-time extracted schema from PreparedModule.
+        // This avoids calling schema() on the already-running WASM instance
+        // where mutable state could produce inconsistent results.
+        if !Self::is_permissive_schema(prepared_schema) {
+            return prepared_schema.clone();
+        }
+
+        self.discovery.clone()
+    }
+}
+
 impl WasmToolWrapper {
    /// Create a new WASM tool wrapper.
    pub fn new(
@@ -486,7 +565,7 @@ impl WasmToolWrapper {
    ) -> Self {
        Self {
            description: prepared.description.clone(),
-            schema: prepared.schema.clone(),
+            schemas: WasmToolSchemas::new(prepared.schema.clone()),
            runtime,
            prepared,
            capabilities,
@@ -504,7 +583,7 @@ impl WasmToolWrapper {

    /// Override the parameter schema.
    pub fn with_schema(mut self, schema: serde_json::Value) -> Self {
-        self.schema = schema;
+        self.schemas = self.schemas.with_override(schema);
        self
    }

@@ -615,9 +694,18 @@ impl WasmToolWrapper {
                }
            })?;

+        // Get typed interface — used for execute.
+        let tool_iface = instance.near_agent_tool();
+
+        // Determine effective schema for type coercion.
+        // Prefer the discovery schema when typed; fall back to the load-time
+        // extracted schema from PreparedModule rather than re-calling the WASM
+        // export on the already-running instance.
+        let effective_schema = self.schemas.effective_for_coercion(&self.prepared.schema);
+
        // Coerce string-encoded values to their schema-declared types.
        // LLMs frequently pass numeric values as strings (e.g. "5" instead of 5).
-        let params = coerce_params_to_schema(params, &self.schema);
+        let params = coerce_params_to_schema(params, &effective_schema);

        // Prepare the request
        let params_json = serde_json::to_string(&params)
@@ -629,7 +717,6 @@ impl WasmToolWrapper {
        };

        // Call execute using the generated typed interface
-        let tool_iface = instance.near_agent_tool();
        let response = tool_iface.call_execute(&mut store, &request).map_err(|e| {
            let error_str = e.to_string();
            if error_str.contains("out of fuel") {
@@ -644,12 +731,13 @@ impl WasmToolWrapper {
        // Get logs from host state
        let logs = store.data_mut().host_state.take_logs();

-        // Check for tool-level error — on failure, call the WASM module's
-        // description() and schema() exports so the LLM can retry with the
-        // correct parameters without us having to include the (large) schema
-        // in every request's tools array.
+        // Check for tool-level error — point the LLM to tool_info for the
+        // full schema instead of dumping ~3.5KB inline.
        if let Some(err) = response.error {
-            let hint = build_tool_hint(tool_iface, &mut store);
+            let hint = format!(
+                "Tip: call tool_info(name: \"{}\", include_schema: true) for the full parameter schema.",
+                self.prepared.name
+            );
            return Err(WasmError::ToolReturnedError { message: err, hint });
        }

@@ -658,47 +746,55 @@ impl WasmToolWrapper {
    }
 }

-/// Maximum characters for the description portion of a tool hint.
-const HINT_DESC_MAX: usize = 500;
-/// Maximum characters for the schema portion of a tool hint.
-const HINT_SCHEMA_MAX: usize = 3000;
+/// Extract metadata (description + schema) from a WASM tool by briefly
+/// instantiating it and calling its `description()` and `schema()` exports.
+/// Analogous to MCP's `list_tools()` — discovers tool capabilities at load time.
+///
+/// Falls back to generic description and permissive schema on failure.
+pub(super) fn extract_wasm_metadata(
+    engine: &wasmtime::Engine,
+    component: &wasmtime::component::Component,
+    limits: &ResourceLimits,
+) -> Result<(String, serde_json::Value), WasmError> {
+    let store_data = StoreData::new(
+        limits.memory_bytes,
+        Capabilities::default(),
+        HashMap::new(),
+        vec![],
+    );
+    let mut store = Store::new(engine, store_data);

-/// Call the WASM module's `description()` and `schema()` exports to build a
-/// hint string.  Returns an empty string if both calls fail or return empty.
-/// Description is capped at [`HINT_DESC_MAX`] chars, schema at
-/// [`HINT_SCHEMA_MAX`] chars.
-fn build_tool_hint(tool_iface: &wit_tool::Guest, store: &mut Store<StoreData>) -> String {
-    let desc = tool_iface
-        .call_description(&mut *store)
+    // Configure fuel + epoch deadline so extraction can't hang
+    if let Err(e) = store.set_fuel(limits.fuel) {
+        tracing::debug!("Fuel not enabled for metadata extraction: {e}");
+    }
+    store.epoch_deadline_trap();
+    let ticks = (limits.timeout.as_millis() / EPOCH_TICK_INTERVAL.as_millis()).max(1) as u64;
+    store.set_epoch_deadline(ticks);
+    store.limiter(|data| &mut data.limiter);
+
+    // Instantiate with minimal linker
+    let mut linker = Linker::new(engine);
+    WasmToolWrapper::add_host_functions(&mut linker)?;
+    let instance = SandboxedTool::instantiate(&mut store, component, &linker)
+        .map_err(|e| WasmError::InstantiationFailed(e.to_string()))?;
+    let tool_iface = instance.near_agent_tool();
+
+    // Extract description (fall back to generic)
+    let description = tool_iface
+        .call_description(&mut store)
+        .unwrap_or_else(|_| "WASM sandboxed tool".to_string());
+
+    // Extract and parse schema (fall back to permissive)
+    let schema = tool_iface
+        .call_schema(&mut store)
        .ok()
-        .unwrap_or_default();
-    let schema = tool_iface.call_schema(&mut *store).ok().unwrap_or_default();
-    if desc.is_empty() && schema.is_empty() {
-        return String::new();
-    }
-    let mut hint = String::new();
-    if !desc.is_empty() {
-        hint.push_str("Description: ");
-        if desc.len() > HINT_DESC_MAX {
-            let end = crate::util::floor_char_boundary(&desc, HINT_DESC_MAX);
-            hint.push_str(&desc[..end]);
-            hint.push('…');
-        } else {
-            hint.push_str(&desc);
-        }
-        hint.push('\n');
-    }
-    if !schema.is_empty() {
-        hint.push_str("Parameters schema: ");
-        if schema.len() > HINT_SCHEMA_MAX {
-            let end = crate::util::floor_char_boundary(&schema, HINT_SCHEMA_MAX);
-            hint.push_str(&schema[..end]);
-            hint.push('…');
-        } else {
-            hint.push_str(&schema);
-        }
-    }
-    hint
+        .and_then(|s| serde_json::from_str::<serde_json::Value>(&s).ok())
+        .unwrap_or_else(|| {
+            serde_json::json!({"type": "object", "properties": {}, "additionalProperties": true})
+        });
+
+    Ok((description, schema))
 }

 #[async_trait]
@@ -712,7 +808,33 @@ impl Tool for WasmToolWrapper {
    }

    fn parameters_schema(&self) -> serde_json::Value {
-        self.schema.clone()
+        self.schemas.advertised()
+    }
+
+    fn discovery_schema(&self) -> serde_json::Value {
+        self.schemas.discovery()
+    }
+
+    /// Compose the tool schema for LLM function calling.
+    ///
+    /// When the advertised schema is permissive (no typed properties), appends
+    /// a hint to the description directing the LLM to call `tool_info` for the
+    /// full parameter schema. This keeps the raw description clean while still
+    /// guiding the LLM.
+    fn schema(&self) -> crate::tools::tool::ToolSchema {
+        let description = if self.schemas.is_advertised_permissive() {
+            format!(
+                "{} (call tool_info(name: \"{}\", include_schema: true) for parameter schema)",
+                self.description, self.prepared.name
+            )
+        } else {
+            self.description.clone()
+        };
+        crate::tools::tool::ToolSchema {
+            name: self.prepared.name.clone(),
+            description,
+            parameters: self.schemas.advertised(),
+        }
    }

    async fn execute(
@@ -749,7 +871,7 @@ impl Tool for WasmToolWrapper {
        let prepared = Arc::clone(&self.prepared);
        let capabilities = self.capabilities.clone();
        let description = self.description.clone();
-        let schema = self.schema.clone();
+        let schemas = self.schemas.clone();
        let credentials = self.credentials.clone();

        // Execute in blocking task with timeout
@@ -759,7 +881,7 @@ impl Tool for WasmToolWrapper {
                prepared,
                capabilities,
                description,
-                schema,
+                schemas,
                credentials,
                secrets_store: None, // Not needed in blocking task
                oauth_refresh: None, // Already used above for pre-refresh
@@ -1232,6 +1354,7 @@ mod tests {
        TEST_GOOGLE_OAUTH_TOKEN, TEST_OAUTH_CLIENT_ID, TEST_OAUTH_CLIENT_SECRET,
        test_secrets_store,
    };
+    use crate::tools::tool::Tool;
    use crate::tools::wasm::capabilities::Capabilities;
    use crate::tools::wasm::runtime::{WasmRuntimeConfig, WasmToolRuntime};

@@ -1246,6 +1369,84 @@ mod tests {
        assert!(runtime.config().fuel_config.enabled);
    }

+    #[tokio::test]
+    async fn test_advertised_schema_stays_permissive_until_sidecar_override() {
+        let discovery_schema = serde_json::json!({
+            "type": "object",
+            "properties": {
+                "query": { "type": "string" },
+                "limit": { "type": "integer" }
+            },
+            "required": ["query"]
+        });
+
+        let runtime = Arc::new(WasmToolRuntime::new(WasmRuntimeConfig::for_testing()).unwrap());
+        let prepared = runtime
+            .prepare("search", b"\0asm\x0d\0\x01\0", None)
+            .await
+            .unwrap();
+        let mut wrapper =
+            super::WasmToolWrapper::new(Arc::clone(&runtime), prepared, Capabilities::default());
+        wrapper.schemas = super::WasmToolSchemas::new(discovery_schema.clone());
+        wrapper.description = "Search documents".to_string();
+
+        // Advertised schema stays permissive; discovery holds the typed schema
+        assert_eq!(
+            wrapper.parameters_schema(),
+            serde_json::json!({
+                "type": "object",
+                "properties": {},
+                "additionalProperties": true
+            })
+        );
+        assert_eq!(wrapper.discovery_schema(), discovery_schema);
+
+        // Raw description is clean — no tool_info hint baked in
+        assert!(!wrapper.description().contains("tool_info"));
+
+        // But schema() composes the hint at display time when advertised is permissive
+        let schema = wrapper.schema();
+        assert!(
+            schema.description.contains("tool_info"),
+            "schema().description should contain tool_info hint: {}",
+            schema.description
+        );
+        assert!(
+            schema.description.contains("include_schema: true"),
+            "hint should mention include_schema: true: {}",
+            schema.description
+        );
+
+        // After sidecar override, both schemas match and hint disappears
+        let wrapper = wrapper.with_schema(serde_json::json!({
+            "type": "object",
+            "properties": {
+                "query": { "type": "string" }
+            },
+            "required": ["query"]
+        }));
+
+        assert_eq!(
+            wrapper.parameters_schema(),
+            serde_json::json!({
+                "type": "object",
+                "properties": {
+                    "query": { "type": "string" }
+                },
+                "required": ["query"]
+            })
+        );
+        assert_eq!(wrapper.discovery_schema(), wrapper.parameters_schema());
+
+        // With typed schema, schema() should NOT include tool_info hint
+        let schema = wrapper.schema();
+        assert!(
+            !schema.description.contains("tool_info"),
+            "schema().description should not contain tool_info hint when typed: {}",
+            schema.description
+        );
+    }
+
    #[test]
    fn test_capabilities_default() {
        let caps = Capabilities::default();
@@ -1788,6 +1989,23 @@ mod tests {
        assert_eq!(result["count"], serde_json::json!("not-a-number"));
    }

+    /// Regression: permissive fallback schema (empty properties) must NOT coerce.
+    /// This documents the bug where WASM tools with no sidecar `parameters` field
+    /// got the permissive fallback, causing coercion to be a no-op and LLM-provided
+    /// string integers to reach the WASM tool un-coerced.
+    #[test]
+    fn test_coerce_noop_with_permissive_schema() {
+        let permissive = serde_json::json!({
+            "type": "object",
+            "properties": {},
+            "additionalProperties": true
+        });
+        let params = serde_json::json!({"query": "test", "count": "10"});
+        let result = super::coerce_params_to_schema(params, &permissive);
+        // With empty properties, no coercion happens — string stays string
+        assert_eq!(result["count"], serde_json::json!("10"));
+    }
+
    /// Regression test: leak scan must run on raw headers (before credential
    /// injection), not after. If it ran post-injection, the host-injected
    /// Slack bot token (`xoxb-...`) would trigger a Block and reject the
@@ -60,12 +60,18 @@ pub trait EmbeddingProvider: Send + Sync {
    }
 }

+/// Default base URL for the OpenAI API.
+const OPENAI_API_BASE_URL: &str = "https://api.openai.com";
+
 /// OpenAI embedding provider using text-embedding-ada-002 or text-embedding-3-small.
+///
+/// Supports any OpenAI-compatible embedding endpoint via [`with_base_url`](Self::with_base_url).
 pub struct OpenAiEmbeddings {
    client: reqwest::Client,
    api_key: String,
    model: String,
    dimension: usize,
+    base_url: String,
 }

 impl OpenAiEmbeddings {
@@ -78,6 +84,7 @@ impl OpenAiEmbeddings {
            api_key: api_key.into(),
            model: "text-embedding-3-small".to_string(),
            dimension: 1536,
+            base_url: OPENAI_API_BASE_URL.to_string(),
        }
    }

@@ -88,6 +95,7 @@ impl OpenAiEmbeddings {
            api_key: api_key.into(),
            model: "text-embedding-ada-002".to_string(),
            dimension: 1536,
+            base_url: OPENAI_API_BASE_URL.to_string(),
        }
    }

@@ -98,6 +106,7 @@ impl OpenAiEmbeddings {
            api_key: api_key.into(),
            model: "text-embedding-3-large".to_string(),
            dimension: 3072,
+            base_url: OPENAI_API_BASE_URL.to_string(),
        }
    }

@@ -112,8 +121,35 @@ impl OpenAiEmbeddings {
            api_key: api_key.into(),
            model: model.into(),
            dimension,
+            base_url: OPENAI_API_BASE_URL.to_string(),
        }
    }
+
+    /// Set a custom base URL for OpenAI-compatible embedding providers.
+    ///
+    /// The URL must use `http://` or `https://` scheme. If no scheme is present,
+    /// `https://` is prepended automatically. Trailing slashes are stripped.
+    pub fn with_base_url(mut self, base_url: &str) -> Self {
+        let url = base_url.trim();
+
+        // Auto-prepend https:// if no scheme is present.
+        let mut url = if !url.starts_with("http://") && !url.starts_with("https://") {
+            tracing::debug!(
+                "No scheme in embedding base URL '{}', prepending https://",
+                url
+            );
+            format!("https://{url}")
+        } else {
+            url.to_string()
+        };
+
+        while url.ends_with('/') {
+            url.pop();
+        }
+
+        self.base_url = url;
+        self
+    }
 }

 #[derive(Debug, Serialize)]
@@ -173,9 +209,11 @@ impl EmbeddingProvider for OpenAiEmbeddings {
            input: texts,
        };

+        let url = format!("{}/v1/embeddings", self.base_url);
+
        let response = self
            .client
-            .post("https://api.openai.com/v1/embeddings")
+            .post(&url)
            .header("Authorization", format!("Bearer {}", self.api_key))
            .json(&request)
            .send()
@@ -575,9 +613,37 @@ mod tests {
        let provider = OpenAiEmbeddings::new("test-key");
        assert_eq!(provider.dimension(), 1536);
        assert_eq!(provider.model_name(), "text-embedding-3-small");
+        assert_eq!(provider.base_url, OPENAI_API_BASE_URL);

        let provider = OpenAiEmbeddings::large("test-key");
        assert_eq!(provider.dimension(), 3072);
        assert_eq!(provider.model_name(), "text-embedding-3-large");
+        assert_eq!(provider.base_url, OPENAI_API_BASE_URL);
+    }
+
+    #[test]
+    fn test_openai_with_base_url_valid() {
+        let provider =
+            OpenAiEmbeddings::new("test-key").with_base_url("https://custom.example.com");
+        assert_eq!(provider.base_url, "https://custom.example.com");
+    }
+
+    #[test]
+    fn test_openai_with_base_url_strips_trailing_slashes() {
+        let provider =
+            OpenAiEmbeddings::new("test-key").with_base_url("https://custom.example.com///");
+        assert_eq!(provider.base_url, "https://custom.example.com");
+    }
+
+    #[test]
+    fn test_openai_with_base_url_http_scheme() {
+        let provider = OpenAiEmbeddings::new("test-key").with_base_url("http://localhost:8080");
+        assert_eq!(provider.base_url, "http://localhost:8080");
+    }
+
+    #[test]
+    fn test_openai_with_base_url_schemeless_prepends_https() {
+        let provider = OpenAiEmbeddings::new("test-key").with_base_url("custom.example.com/v1");
+        assert_eq!(provider.base_url, "https://custom.example.com/v1");
    }
 }
@@ -55,7 +55,9 @@ pub use embeddings::{
 };
 #[cfg(feature = "postgres")]
 pub use repository::Repository;
-pub use search::{RankedResult, SearchConfig, SearchResult, reciprocal_rank_fusion};
+pub use search::{
+    FusionStrategy, RankedResult, SearchConfig, SearchResult, fuse_results, reciprocal_rank_fusion,
+};

 use std::sync::Arc;

@@ -332,6 +334,8 @@ pub struct Workspace {
    storage: WorkspaceStorage,
    /// Embedding provider for semantic search.
    embeddings: Option<Arc<dyn EmbeddingProvider>>,
+    /// Default search configuration applied to all queries.
+    search_defaults: SearchConfig,
 }

 impl Workspace {
@@ -343,6 +347,7 @@ impl Workspace {
            agent_id: None,
            storage: WorkspaceStorage::Repo(Repository::new(pool)),
            embeddings: None,
+            search_defaults: SearchConfig::default(),
        }
    }

@@ -355,6 +360,7 @@ impl Workspace {
            agent_id: None,
            storage: WorkspaceStorage::Db(db),
            embeddings: None,
+            search_defaults: SearchConfig::default(),
        }
    }

@@ -370,6 +376,16 @@ impl Workspace {
        self
    }

+    /// Set the default search configuration from workspace search config.
+    pub fn with_search_config(mut self, config: &crate::config::WorkspaceSearchConfig) -> Self {
+        self.search_defaults = SearchConfig::default()
+            .with_fusion_strategy(config.fusion_strategy)
+            .with_rrf_k(config.rrf_k)
+            .with_fts_weight(config.fts_weight)
+            .with_vector_weight(config.vector_weight);
+        self
+    }
+
    /// Get the user ID.
    pub fn user_id(&self) -> &str {
        &self.user_id
@@ -709,13 +725,13 @@ impl Workspace {
    /// Hybrid search across all memory documents.
    ///
    /// Combines full-text search (BM25) with semantic search (vector similarity)
-    /// using Reciprocal Rank Fusion (RRF).
+    /// using the configured fusion strategy.
    pub async fn search(
        &self,
        query: &str,
        limit: usize,
    ) -> Result<Vec<SearchResult>, WorkspaceError> {
-        self.search_with_config(query, SearchConfig::default().with_limit(limit))
+        self.search_with_config(query, self.search_defaults.clone().with_limit(limit))
            .await
    }

@@ -12,7 +12,7 @@ use uuid::Uuid;
 use crate::error::WorkspaceError;

 use crate::workspace::document::{MemoryChunk, MemoryDocument, WorkspaceEntry};
-use crate::workspace::search::{RankedResult, SearchConfig, SearchResult, reciprocal_rank_fusion};
+use crate::workspace::search::{RankedResult, SearchConfig, SearchResult, fuse_results};

 /// Database repository for workspace operations.
 pub struct Repository {
@@ -415,7 +415,7 @@ impl Repository {
            Vec::new()
        };

-        Ok(reciprocal_rank_fusion(fts_results, vector_results, config))
+        Ok(fuse_results(fts_results, vector_results, config))
    }

    /// Full-text search using PostgreSQL ts_rank_cd.
--- a/Show More
+++ b/Show More