* refactor: encapsulate leaked abstractions from main.rs and app.rs into owning modules
Move module-specific initialization logic out of main.rs (1222→665 lines, -46%) and
app.rs (944→780 lines, -17%) into their respective owning modules as public factory
functions. This enforces separation of concerns so that adding a new DB backend, MCP
transport, or channel doesn't require editing main.rs/app.rs.
Key changes:
- Tracing init functions → src/tracing_fmt.rs
- DB connection factory (connect_with_handles + DatabaseHandles) → src/db/mod.rs
- Secrets store factory (create_secrets_store) → src/secrets/mod.rs
- MCP transport dispatch factory (create_client_from_config) → src/tools/mcp/factory.rs
- Orchestrator setup (setup_orchestrator + OrchestratorSetup) → src/orchestrator/mod.rs
- WASM channel setup (setup_wasm_channels) → src/channels/wasm/setup.rs
- Worker entry points (run_worker, run_claude_bridge) → src/worker/mod.rs
- Shared CLI secrets init (init_secrets_store) → src/cli/mod.rs
- Tunnel startup (start_managed_tunnel) → src/tunnel/mod.rs
- Onboard check (check_onboard_needed) → src/setup/mod.rs
- ExtensionManager unified MCP: uses create_client_from_config via McpProcessManager,
enabling stdio/Unix transports for hot-activated MCP servers
- Deduplicated ~130 lines of secrets store init across cli/mcp.rs and cli/tool.rs
- CLAUDE.md updated with module-owned initialization guideline
[skip-regression-check]
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* refactor: address review feedback — deduplicate db factory, extract channel helper
- connect_from_config() now delegates to connect_with_handles() to eliminate
duplicated backend-matching logic (Copilot review feedback)
- Extract register_channel() helper from setup_wasm_channels() loop body
to improve readability (Gemini review feedback)
[skip-regression-check]
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* style: fix rustfmt line wrapping in setup_wasm_channels
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* test: add integration test for module-owned initialization factories
Exercises the full factory chain end-to-end to verify nothing was lost
when initialization logic was moved from main.rs/app.rs into owning modules:
- connect_with_handles returns Database + populated backend handles
- connect_from_config delegates correctly (produces working Database)
- secrets::create_secrets_store builds working store from DatabaseHandles
- db::create_secrets_store standalone factory round-trips secrets
- Both secrets factories produce compatible stores (cross-read works)
- ExtensionManager constructs with McpProcessManager and is functional
- DatabaseHandles default is empty
All tests run without external services using libsql in-memory/tempfile.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* fix: wire cli/mcp.rs and cli/tool.rs to shared init_secrets_store()
Both files had inline implementations identical to cli::init_secrets_store().
Replace with delegation to complete the claimed deduplication.
[skip-regression-check]
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* style: fix rustfmt line wrapping in integration test
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* fix(review): remove unused Config import and deduplicate Error Handling section
- Remove `#[allow(unused_imports)]` and unused `use crate::config::Config`
from cli/tool.rs (no longer needed after delegating to shared
`cli::init_secrets_store()`)
- Remove duplicate Error Handling subsection from CLAUDE.md Key Patterns
(all four bullets already exist in Code Style section and
review-discipline.md)
Addresses Copilot review comments.
[skip-regression-check]
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* fix(review): address remaining Copilot review comments
- secrets/mod.rs: clarify docstring that None is a normal no-db condition
- app.rs: add comment explaining the empty_handles fallback path
- orchestrator/mod.rs: combine duplicated sandbox condition into single block
- setup/mod.rs: document env var reads and thread-safety caveat
[skip-regression-check]
Co-Authored-By: Claude Opus 4.6 <[email protected]>
---------
Co-authored-by: Claude Opus 4.6 <[email protected]>
Co-authored-by: Henry Park <[email protected]>
* feat: Move debug log truncation from agent loop to REPL channel
Full tool output now flows through StatusUpdate so the web gateway
gets untruncated content. The REPL channel truncates at display time
(200 chars for tool results, thinking, and status messages).
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* feat: truncating fmt layer for terminal, full logs for web gateway
Instead of truncating debug output at each LLM call site (fragile),
use a custom MakeWriter on the fmt layer that caps each tracing event
at 500 bytes before flushing to stderr. The web gateway WebLogLayer
still receives full untruncated content for /api/logs/events SSE.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
* fix: UTF-8 safe truncation in truncate_for_preview, remove double truncation
- Use char_indices() instead of byte-based slicing to find the cut
point, preventing panics on multi-byte characters (emoji, CJK, etc.)
- Remove redundant truncation in REPL channel (agent loop already
truncates ToolResult previews to 200 chars)
- Add 9 unit tests covering edge cases: empty, exact length, multi-byte
UTF-8 (emoji, CJK), mixed scripts, newline collapsing, whitespace
Addresses PR #65 review comments.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
---------
Co-authored-by: Claude Opus 4.6 <[email protected]>
