sync-memorieslisted

# Sync Memories — Session Knowledge Capture Update persistent memory systems with session accomplishments so future sessions have accurate context. ## When to use - After completing a PR / feature / refactor session - After a release (version bump, changelog, tag) - After discovering important architecture or gotcha info - When existing memories are stale (test counts, versions, READMEs outdated) ## Workflow ### 1. Gather current state ```bash git log --oneline -10 git diff --stat HEAD~3 2>/dev/null || git diff --stat HEAD git branch --show-current git status # If JS/TS project: node -p "require('./package.json').version" 2>/dev/null npm test 2>&1 | grep -E "Tests:|Test Suites:" | tail -2 ``` ### 2. Pick the right memory system | Tool present | Where memories live | Update via | |---|---|---| | Claude Code project memory | `~/.claude/projects/<slug>/memory/*.md` | direct file edit + MEMORY.md pointer line | | Serena MCP (`.serena/`) | per-project memories | `serena.write_memory(name, content)` | | claude-mem MCP | FTS5 DB | `save_memory({project, title, text})` | | `.agents/memory/<project>.md` | tracked in repo | append a `## Session YYYY-MM-DD` block | Use all that apply. They don't conflict; recall queries each separately. ### 3. Update — one fact per file Each memory file holds one fact. Frontmatter required (see global CLAUDE.md). For project memories: `type: project`, link related notes with `[[name]]`, and add a one-line pointer to `MEMORY.md` so the index l