sdlc-context

# sdlc-context The read side of the knowledge graph. Project memory grows past the context window; this pulls the *relevant slice* in before the work starts — retrieval by relevance, not recency — so the SDLC begins informed by history instead of re-deriving (or contradicting) what was already decided. **Gate — only acts when the knowledge graph is enabled and built:** `python3 "${CLAUDE_SKILL_DIR}/../sdlc-kg/scripts/kg.py" status .sdlc` - *disabled* → skip silently; run the SDLC without a brief. - enabled but *graph: not built* → build it first (`/sdlc-kg`), or skip this run. When ready, assemble a **Context Brief** for the goal at hand (its text + the files/components it will touch are the query seeds): 1. **Query the graph** for prior art: `graphify query "<the goal, plus the modules/files it will touch>"` — BFS for surrounding context, `--dfs` to trace one chain. If the **graphify MCP** is connected, prefer the live `query_graph` tool. Quote each answer's `source_location`; if the graph lacks something, say so — never invent an edge. **When the graph comes up empty** (not found / low-confidence), log the unanswered question as a gap so the KG tracks what it doesn't know yet: `python3 "${CLAUDE_SKILL_DIR}/../sdlc-kg/scripts/kg.py" gap log "<the unanswered question>" .sdlc` (deduped + fail-open; it becomes the backlog the loop can later fill). Review the running list with `kg.py gap list .sdlc`. 2. **Recall past decisions/findings** on those comp...