research-wiki

You can track research relationships by initializing a project-level wiki that stores papers, ideas, and experiments while maintaining typed edges in a graph to map their connections. This persistent knowledge base automatically renders relationship views on entity pages.

Compiling arXiv literature is handled by ingesting papers with arXiv metadata or backfilling syncs via arXiv IDs. The wiki stores these papers as entity pages and logs all updates in an append-only timeline log.

To summarize research context within token limits, the wiki generates a bounded query_pack up to 8000 characters. This pack summarizes top gaps, clusters, failed ideas, and active chains for rapid review.

Yes, you can validate the knowledge base structure using linting operations and monitor coverage with built-in stats. These operational lifecycle features ensure your wiki artifacts remain well-formed and comprehensive.

Yes, the research wiki operates with no external dependencies. It relies on a root-level SKILL.md entrypoint and a canonical helper script resolution chain to write wiki artifacts directly to your project directory.

Initializing a project research wiki generates artifacts in a dedicated directory, including index.md, log.md, graph/edges.jsonl, and individual entity pages. These files persistently store your research data and relationship graphs.