sync-context

Documentation drift occurs when code changes leave context files like AGENTS.md and CLAUDE.md outdated. You detect documentation drift by running automated drift checks that scan the docs/ai layer, compare it against recent code changes, and report misalignments.

To keep AGENTS.md and CLAUDE.md in sync after code changes, apply automated drift detection checks to the repository. This generates targeted proposals with old_text to new_text edits, which you review and explicitly approve before applying any updates.

No, you cannot automatically apply updates without review. The governance workflow enforces a read-then-propose model, requiring explicit approvals before applying any changes to AGENTS.md, CLAUDE.md, or docs/ai files to ensure up-to-date and accurate context.

The best way to maintain AI context files during periodic reviews is running drift detection across the repository's AI context layer. This process identifies misalignment and proposes focused updates to keep documentation current and reduce drift over time.

Yes, documentation drift detection works without external dependencies. It operates independently to scan the repository's context layer, detect misalignment across AGENTS.md and CLAUDE.md, and propose targeted edits for explicit approval.