What problem does it solve? Brownfield projects lack a reliable, graph-native understanding of their own codebase, so downstream planning and navigation workflows operate on guesswork. This Skill enumerates the full project file universe, classifies files by value tier, deep-scans the highest-value evidence, and publishes provisional nodes, edges, and coverage artifacts that a later build step turns into a project cognition graph. ## Core Features & Use Cases - Value-weighted repository inventory: Resolves the candidate scan set through the project-cognition runtime, applies .cognitionignore rules, and classifies every path into P0-P3 value tiers with explicit dispositions. - Subagent-orchestrated deep scanning: Dispatches bounded MapScanPacket lanes to subagents with machine-checkable worker-result handoffs, coverage ledgers, and blocked-state recovery contracts. - Provisional graph artifacts: Emits nodes.json, edges.json, observations.json, coverage.json, and workbench state files under .specify/project-cognition/ for handoff to /sp-map-build. - Use Case: Before running graph-based navigation or planning in a spec-kit project, invoke this Skill to produce a complete evidence baseline so /sp-map-build can reconstruct the cognition graph from live repository surfaces. ## Quick Start Ask the agent to run /sp-map-scan on this repository, optionally naming a subsystem or directory to emphasize, so it produces the project-cognition evidence baseline.
