Map

Build or refresh the second brain's map of this codebase.

Current state

• Brain: !ls .neo/brain 2>/dev/null || echo "MISSING"
• ARCHITECTURE.md size: !wc -l .neo/brain/ARCHITECTURE.md 2>/dev/null || echo "n/a"
• Graphify: !command -v graphify >/dev/null 2>&1 && echo "available" || echo "not installed"

Steps

1. Guard. No .neo/brain/ → stop and suggest /neo:init first. This command fills the map; init builds the shelf.

2. Recon fan-out. Spawn keymakers IN PARALLEL (one message, 3–4 spawns), each with the full 6-section contract and ONE angle:

   • Stack & entry points: languages, frameworks + versions from manifests, build/test/lint commands, where execution starts.
   • Layout & boundaries: top-level modules, what owns what, how layers talk (imports, APIs, events).
   • Patterns & conventions: error handling, naming, state management, test conventions — concrete one-liners with file:line examples.
   • Gotchas & hotspots: the biggest files, circular-looking dependencies, anything where "X looks like Y but is actually Z".

   If graphify is available and a graph.json exists, tell each keymaker to prefer graphify query/path/explain over broad grep.

3. Synthesize via shadow. Spawn neo-shadow with ALL recon findings as the delta. Contract: rewrite ARCHITECTURE.md (Stack / Layout / Patterns / Gotchas — concrete, file-anchored, no prose padding); extract any subsystem too big for a section into .neo/brain/topics/<name>.md and [[wiki-link]] it; update INDEX.md pointers; touch nothing else.

4. Report. Tell the user what the map now covers, its size (flag if ARCHITECTURE.md pushed past ~500 lines — shadow should have split topics), and anything the recon flagged as unclear or contradictory.

Re-running is safe: shadow rewrites from current findings, and prior knowledge lives in git history.