What problem does it solve?
This Skill helps you ask questions about an EmpiricalWiki knowledge base and receive evidence-backed answers that reference the exact wiki pages, optionally crystallizing the result back into new or updated wiki content.
Core Features & Use Cases
- Global-context aware retrieval: Loads
wiki/graph/context_brief.md and wiki/graph/open_questions.md, then uses wiki/index.md to locate the most relevant pages.
- Citation-grounded synthesis: Produces structured answers where key claims are linked using
[[slug]] wikilinks, and flags uncertainty when the wiki lacks sufficient evidence.
- Knowledge compounding (crystallize): In
--crystallize mode, writes query results to wiki/outputs/, and may create or update wiki/concepts/ or wiki/claims/ plus append graph edges and rebuild derived context.
- Evidence & graph integration: Updates
wiki/graph/edges.jsonl, wiki/index.md, and wiki/log.md so future questions benefit from the new relationships.
Quick Start
Ask a research question and let the skill retrieve, synthesize, and cite relevant wiki pages: run /ask with your question, for example: "LoRA 和 Adapter 的核心区别是什么?".