update-project-knowledge

To update CLAUDE.md and agent memories after a code change, run a knowledge synchronization process at the end of your implementation task. This gathers context, identifies documentation gaps, and updates relevant knowledge files to reflect new conventions, packages, or patterns.

Keeping project documentation aligned with new code patterns is best achieved through automated knowledge synchronization. It updates CLAUDE.md and subagent MD files after major implementation tasks, ensuring documentation stays aligned with reality for faster onboarding.

You need to synchronize your knowledge base during a development workflow at the end of a significant implementation task. This is required when new conventions, packages, resources, or patterns are discovered and need to be reflected in CLAUDE.md and agent memories.

Yes, updating agent memories works for subagent MD files as well. The synchronization process updates CLAUDE.md, agent memories, and subagent MD files across code changes, new conventions, packages, resources, or patterns discovered.

Project knowledge drifts after major implementation tasks because documentation is not automatically updated with new conventions or patterns. Running a knowledge synchronization process after significant work reduces drift and miscommunication by aligning documentation with reality.