sdlc-kg

# sdlc-kg **Optional. Only acts when `.sdlc/config.json` → `knowledge_graph.enabled` is `true`.** The graph accumulates two things, to keep enhancing the project's learnings: - **External research** — auto-captured from every `WebSearch` / `WebFetch` into `.sdlc/knowledge/research/web/` (the `research_capture` hook does this whenever KG is enabled). - **Internal analysis** — durable findings + lessons you write as markdown to `.sdlc/knowledge/analysis/` during Research / Plan-Review / Retrospective. At `scope: full` the **code** is graphed too; at `scope: research` the code is skipped. 1. **Check status:** `python3 "${CLAUDE_SKILL_DIR}/scripts/kg.py" status .sdlc` If it reports *disabled*, tell the user how to turn it on (set `knowledge_graph.enabled: true`; optionally `scope: "research"` to skip code) and stop. 2. **Get the plan:** `python3 "${CLAUDE_SKILL_DIR}/scripts/kg.py" plan .sdlc .` It prints the corpus path, whether the code is included, and the builder. 3. **Build / refresh** with the builder (default `graphify`) over the planned inputs: - `scope: research` → run the builder on `.sdlc/knowledge/` only. - `scope: full` → run the builder on `.sdlc/knowledge/` **and** the repo code, merged into one graph. - On re-runs use the builder's incremental update. For graphify you may invoke the `/graphify` skill on those paths, or the `graphify` CLI directly. - If the builder isn't installed, say so (for graphify: `pip install graphifyy`) and ...