What problem does it solve?
Ensures documentation and session memories do not drift from the codebase after feature work, migrations, refactors, or bug fixes by verifying, repairing, and committing only documentation and memory updates while avoiding staging unrelated code changes.
Core Features & Use Cases
- Rebuild code-review graph: incrementally update or note missing graph state to keep review metadata accurate.
- Reflect on session learnings: detect new modules, migrations, tools, or architecture changes to target checks.
- CLAUDE.md and roadmap verification: confirm version, architecture summary, build and lint commands remain accurate.
- Serena memory sync: read and apply targeted edits to five memories (project_overview, project_structure, style_and_conventions, suggested_commands, task_completion_checklist) using filesystem ground truth.
- Safe commit workflow: run formatting checks, stage only doc/memory edits, and commit with a standardized message; skip or note missing surfaces gracefully.
- Use case example: after finishing a feature and adding a migration, run the skill to update CLAUDE.md version references and Serena memories, then commit the drift fixes in a single docs commit.
Quick Start
Invoke the session-sync skill at the end of your work session with an optional scope hint, for example: /session-sync after adding WebSocket adapter + migration 043.
