Sevenbelow

Three-way gap analysis: canonical specs vs vault wiki vs actual code. Finds missing features, claimed-but-broken, and undocumented implementations.

Build the knowledge vault from configured sources. Runs the ETL pipeline: survey sources, preview extraction, execute parallel build, generate Graphify knowledge graph. Supports building all sources or specific ones (e.g., `library:build jira` or `library:build axon_bridge,specs`).

Capture session state at the end of a work session. Writes structured checkpoint file, updates memory, and comments on PM tasks with progress.

Compile wiki articles from vault sources. The Librarian's core job: raw sources become structured wiki articles tagged with [VERIFY], [CONFLICT], [PLANNED].

Set up and configure The Library. Run on first use or to update settings. Handles Reading Room setup, vault initialization, PM provider selection, Graphify toggle, branding, and MCP registration.

Ingest source material into the vault. Single mode for one source, batch mode (--batch) for multiple sources during setup or bulk import.

Memory lifecycle management. Prune stale entries, validate references, aggregate related memories, detect conflicts, track age. Garbage collection for the knowledge layer.

Convert a spec or design document into PM epics and tasks. Parses sections, extracts deliverables, maps to project hierarchy.

Ask the Librarian questions. Routes to vault parsing, Graphify graph queries, or PM state depending on the question. Use --report for formatted status summary.

Validate a specific completion claim against running code. Verdict: VERIFIED, PARTIAL, or FAILED — with evidence. Enforces evidence before claims.

Sync project state from Jira or Linear. Reports open, stale, blocked, and recently closed tasks. Compares PM state against vault tags.

Scan vault for [VERIFY], [CONFLICT], [PLANNED] tags and create PM tasks from them. Deduplicates against existing tasks.