mirror of
https://github.com/RooVetGit/Roo-Code.git
synced 2026-09-17 23:51:08 +00:00
- Added SelectActiveIntentTool for handshake protocol - Implemented middleware gatekeeper in tool execution pipeline - Integrated RecordAgentTraceTool for SHA-256 content hashing - Created .orchestration/ artifacts for machine-readable audit trails
1.9 KiB
1.9 KiB
Hooks Directory - Clean Hook Interfaces
This directory contains the clean, composable hook implementations for the governed AI-native IDE.
Pre-Hooks (Execution Phase)
| Hook | Phase | Purpose |
|---|---|---|
validateIntent |
Phase 1 | Ensures active intent exists before execution |
classifyCommand |
Phase 2 | Categorizes SAFE vs DESTRUCTIVE commands |
enforceScope |
Phase 2 | Validates file against intent's owned_scope |
authorizeHITL |
Phase 2 | Shows user approval dialog for destructive commands |
loadIntentContext |
Phase 1 | Loads intent context from active_intents.yaml |
Post-Hooks (Completion Phase)
| Hook | Phase | Purpose |
|---|---|---|
hashContent |
Phase 3 | Generates SHA-256 content hash for traceability |
classifyMutation |
Phase 3 | Determines REFACTOR vs EVOLUTION |
recordTrace |
Phase 3 | Appends trace entry to agent_trace.jsonl |
validateLock |
Phase 4 | Checks for stale files before write |
captureInitialHash |
Phase 4 | Captures pre-write hash for optimistic locking |
Integration Points
These hooks are wired into the following locations:
- Pre-execution:
src/core/assistant-message/presentAssistantMessage.ts - Post-execution:
src/core/tools/WriteToFileTool.ts - Pre-prompt:
src/core/task/Task.ts
Architecture Philosophy
- Isolated: Each hook is independent and focused
- Composable: Hooks can be combined in different orders
- Fail-safe: All hooks are wrapped in try/catch blocks
- Non-intrusive: Added without modifying core logic