AI coding agent regression memory - Anchor Docs

Help AI coding agents avoid repeating known regressions by retrieving revert, rollback, root-cause, hotfix, and incident evidence before edits.

A coding agent may make a change that looks correct locally but repeats an old production regression or rejected approach.

• Extracts regression memory from PR titles, bodies, labels, comments, review summaries, and commit messages.
• Ranks regression evidence strongly when target files, symbols, or API contracts match the current task.
• Adds strict mode so weak or stale matches do not over-guide risky changes.

anchor review --strict
anchor explain src/auth/access.ts
anchor org impact --org my-org --strict

Anchor docs · GitHub