What problem does it solve? Keeping a research wiki current requires many coordinated steps — gathering new sources, routing them to owners, ingesting content, checking transcripts, linting, fixing, and reporting — and doing them by hand is error-prone and hard to resume after interruption. ## Core Features & Use Cases - End-to-end cycle orchestration: Chains discovery, triage, ingest, checker, lint, claims, synthesis, refresh, report, and a single commit into one command with quick, full, ingest-only, discover-only, and prompt-for-urls modes. - Resumable runs: Every step writes JSON and Markdown outputs to a run folder, and a scratchpad lets an interrupted cycle resume from the last completed phase via --resume. - Human review gates: Pauses after discovery for approval and stages entries in _inbox/proposed/ for a morning review before promotion into the wiki. - Use Case: Run /wiki-cycle --full weekly to gather queued sources, ingest them with parallel workers, verify long transcripts with a checker agent, update claims and best-practices pages, and receive a consolidated run report. ## Quick Start Ask the AI to run the wiki cycle on the current notebook to gather, ingest, lint, and report on new research sources.
