Review completed local project sessions for reusable lessons, verify them against project evidence, remove duplicate candidates and prepare exact proposals for owner approval. Use for manual session analysis; this skill does not schedule scans or save…
Use AFTER research-build (or manual addition of new knowledge) to propose updates to the Obsidian vault under brain/concepts/. Activates POST-TASK only — not mid-session, not parallel. Triggers: 'update the vault', 'sync brain', 'ingest into concepts', 'add…
Consult a connected research-brain MCP vault for source-backed context during project research; verify local applicability, surface freshness and gaps, and keep saves behind explicit approval.
User-facing CRUD over the author-monitoring roster (config/authors.yaml). One intent per turn: ADD an author the user has decided to follow, LIST the roster, PAUSE / RESUME one, or REMOVE one. Every operation goes through scripts/manage_authors.py — the only…
Answer READ-ONLY status, findings & HEALTH questions about THIS system's own state — what it monitors, on what cadence, what digests it has produced, and whether everything is healthy and correctly set up (config valid, feeds alive, delivery wired, digest…
Adversarial, multi-agent CONTENT-TRUST audit of the standing knowledge vault — re-verify existing brain/concepts/ against their own sources: to catch hallucinations, the single-source illusion, staleness, contradictions, overconfidence and tag drift. NOT…
Present a friendly, plain-language menu of what this system can do, grouped by the user's goal, with concrete example sentences they can say. This is an ORIENTATION skill — it does NOT do the work itself; it shows the user what's possible and routes them to…
User-facing CRUD over the concept vault (brain/concepts/). One intent per turn: CREATE a new monitored concept via a short interview (name → sources → keywords/tags → run research → fill the note), LIST concepts, UPDATE (re-research) one, or ARCHIVE / RESTORE…