graphifylisted

# /graphify (Codeflare-managed skill) This skill drives `/graphify` knowledge-graph extraction inside the Codeflare container. The `graphifyy` Python tool is pre-installed at build time (`uv tool install graphifyy[mcp,sql,pdf]`); the `graphify` CLI is on PATH at `/root/.local/bin/graphify` and the MCP server is pre-registered in `~/.claude.json` as `/root/.local/share/uv/tools/graphifyy/bin/python -m graphify.serve`. You do not need to install or wire anything. ## Quick Reference **Python interpreter:** All `python3` code blocks in this skill must use graphify's own interpreter: `/root/.local/share/uv/tools/graphifyy/bin/python`. System `python3` cannot import graphify. The CLI at `/root/.local/bin/graphify` is on PATH. ### Recipe 1: AST-only update (free, no tokens) ``` bash /home/user/.claude/plugins/graphify/scripts/safe-graphify-update.sh . ``` Re-extracts code structure only. Use after source code changes. Memory-safe (OOM-guarded). ### Recipe 2: Full semantic update (existing repo) 1. Detect files (Step 1 below) 2. Check semantic cache (Step B0) - note cached vs uncached counts 3. Present AskUserQuestion: AST-only vs Full (Note 8). Include subagent count: `ceil(uncached_doc_paper_files / 22) + uncached_image_count` (images get own chunk) 4. If Full chosen: start AST (Part A) in background, dispatch first semantic wave in parallel 5. Split uncached non-code files into chunks of 22 (images get own chunk) 6. Dispatch waves of at most 10 Sonnet subagents (Note 9: `mode