| name | repo-review |
| description | Review an existing repository for architecture issues, runtime risks, testing gaps, and code quality regressions. Use when asked to review, audit, harden, or check whether the current implementation is stable. |
| metadata | {"ccbot":{"emoji":"🧪"}} |
Repo Review
Use this skill when the task is to review an existing codebase rather than build a new feature from scratch.
Review order
- Read the current entrypoints and runtime boundaries.
- Inspect the main orchestration path before edge modules.
- Check whether docs, prompts, and runtime behavior still match.
- Run the smallest useful validation first, then broader tests.
- Report root-cause issues, not just symptoms.
Focus points for ccbot-style agent runtimes
Channel -> AgentTeam -> Supervisor -> WorkerPool -> Worker
- prompt / preset /
.claude/settings.json alignment
- control commands and runtime observability
- scheduler / memory / workspace persistence
- channel progress, final reply, and worker-result semantics
Good review output
Always summarize:
- what is already solid
- what is misleading or partially wired
- what can break at runtime
- what should be fixed now vs later
Validation
Prefer:
uv run ruff check .
uv run pytest
If testing a live runtime path, start with CLI before bot channels.