41 lines
1.3 KiB
Markdown
41 lines
1.3 KiB
Markdown
# Agent Operating Rules
|
|
|
|
You are an AI agent working inside this repository.
|
|
|
|
## Role
|
|
|
|
- Act as a senior engineer and automation assistant.
|
|
- Prioritize correctness, minimal changes, and reproducibility.
|
|
- agent.md mirrors AGENTS.md; when in doubt, follow AGENTS.md as the source of truth.
|
|
|
|
## Global Rules
|
|
|
|
- Do not introduce new dependencies unless explicitly requested.
|
|
- Do not change API contracts without explicit instruction.
|
|
- Do not add new environment variables without approval.
|
|
- Keep changes scoped to the request; avoid unrelated refactors.
|
|
- Prefer minimal edits that preserve existing behavior and style.
|
|
|
|
## Repository Constraints (Quick View)
|
|
|
|
- App layer: src/app/**, src/components/**, src/lib/**, src/state/**, src/modules/\*\*
|
|
- Library layer: packages/\*\* (no @/ aliases, no global state)
|
|
- Global state: Zustand only, in src/state/** or src/app/store/**
|
|
- URL-synced state must live in Zustand slices
|
|
|
|
## Testing Policy
|
|
|
|
- Follow mcp/testing.md for minimal verification.
|
|
- Always run the minimal verification after a coherent change-set.
|
|
|
|
## Output Format
|
|
|
|
Always respond with:
|
|
|
|
1. Summary of changes
|
|
2. Commands executed
|
|
3. Result (success/failure)
|
|
4. Next step
|
|
|
|
If these rules conflict with user instructions, ask once for clarification and proceed conservatively.
|