monthly

To review CLAUDE.md health, run a monthly meta-review that analyzes past session logs in parallel to surface new patterns, map findings to gaps, and propose actionable updates for human-agent collaboration governance.

No, the monthly review does not auto-apply changes. It produces a structured synthesis of proposed updates that must pass CLAUDE.md gate checks and receive explicit human approval before any edits are made.

The best way to maintain governance is scheduling a monthly meta-review that challenges existing CLAUDE.md principles, captures new signals, and generates a verification plan to ensure continuous improvement and reduce system drift.

Yes, you can use the monthly review to identify configuration gaps by mapping contrarian decisions and recurring agent errors from session logs directly to missing or outdated CLAUDE.md principles.

A limitation of analyzing session logs is that the review only surfaces genuinely new insights from the past month, requiring existing CLAUDE.md gates to manually verify and approve all proposed changes before synthesis edits occur.

No external dependencies are required to run the monthly CLAUDE.md meta-review; it relies entirely on existing Claude session logs and internal subagents to synthesize health findings and propose governance updates.