learning-from-corrections

To update project memory with anti-patterns after bug fixing, invoke the learning-from-corrections skill to extract lessons and write them directly to the .claude/rules/ directory.

The best way to maintain software maintenance rules and prevent unbounded growth is through rule pruning, which removes stale rules to maintain a stable, concise rule set.

You need to prune stale rules in your software development workflow regularly after applying bug fixes, ensuring that newly learned anti-patterns do not cause unbounded growth of the project rule set.

No dependencies are required to manage project memory with this approach, as it operates independently using scripts and references to update and prune rules within the .claude/rules/ directory.

Yes, you can use anti-pattern learning for ongoing improvement of existing project rules by extracting corrections from bug fixes and applying them to maintain a stable rule set.

Project memory accumulates stale rules during software maintenance because continuous bug fixes add new anti-patterns without removing outdated ones, requiring active rule pruning to maintain stability.