graphifylisted

# Graphify Build a compact JSON knowledge graph for a software project. Result: `knowledge_graph.json` + per-cluster `.md` files readable in ~70x fewer tokens than scanning source files directly. ## When to Use - New project — no graph exists yet - Project has changed significantly — graph is stale - User says "graphify", "build graph", "граф на проекта" ## Step 1 — Initialize Output Directory ```bash python skills/graphify/scripts/init_graph.py "<ProjectName>" "<{{WIKI_PATH}}/Project/graph>" ``` This creates placeholder files. Proceed to fill them in. ## Step 2 — Discover Project Structure Read the entry point first (80% of architecture info lives here): ```bash # Python/FastAPI head -120 backend/main.py # app.include_router() calls -> all domains grep "include_router" backend/main.py # full list of routers # Node.js/Express head -80 src/app.js # app.use() calls -> all route domains grep "app.use\|require.*router" src/app.js # Then list top-level directories only ls backend/ ls src/ ``` Do NOT read all files — entry point + 2-3 `ls` commands are enough for Step 2. ## Step 3 — Identify Clusters From the entry point, group related routers/modules into 6-14 domain clusters. **Cluster naming (use these standard names when they fit):** `auth` | `invoicing` | `expenses` | `accounting` | `tax_compliance` | `hr_payroll` | `clients` | `platform` | `ai` | `frontend` | `core` | `api` | `ml` | `tasks` | `queue_workers` | `crawler` | `graph` | `ide