swarmvault

name	swarmvault
description	SwarmVault graph-first workflow. Use to read the compiled wiki and query the knowledge graph before broad file search.

SwarmVault

SwarmVault compiles curated sources in raw/ into a queryable wiki in wiki/ and a knowledge graph in state/graph.json.

Rules

• Read swarmvault.schema.md before compile or query style work. It is the canonical schema path.
• Treat raw/ as immutable source input.
• Treat wiki/ as generated markdown owned by the agent and compiler workflow.
• If SWARMVAULT_OUT is set, resolve generated artifact paths like raw/, wiki/, and state/ under that directory.
• Read wiki/graph/report.md before broad file searching when it exists; otherwise start with wiki/index.md.
• For code and graph questions (where is X, what calls Y, structure, impact), prefer swarmvault graph query, swarmvault graph path, and swarmvault graph explain over broad grep/glob searching; read source files directly only when editing them or when the graph lacks detail.
• Preserve frontmatter fields including page_id, source_ids, node_ids, freshness, and source_hashes.
• When asked for durable research, reviews, or handoff artifacts, save the answer into wiki/outputs/; answer quick questions directly in chat without writing files.
• Prefer swarmvault ingest, swarmvault compile, swarmvault query, and swarmvault lint for SwarmVault maintenance tasks.

Entry points

• swarmvault ingest <path> — register a new source
• swarmvault compile — refresh wiki pages and graph
• swarmvault query "<question>" — save-first multi-step query
• swarmvault graph query|path|explain — deterministic graph traversal
• swarmvault lint — wiki health and contradiction checks