ask

Answers questions against a research wiki and optionally crystallizes synthesized answers back into wiki pages.

Updated Jul 15, 2026
One-click install
npx skills add https://github.com/1549171506/shizhengmubangai1 --skill ask-1549171506
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ask
Source: https://github.com/1549171506/shizhengmubangai1/tree/main/Desktop/123/empiricalwiki/i18n/en/skills/ask
Command: npx skills add https://github.com/1549171506/shizhengmubangai1 --skill ask-1549171506

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Querying a growing personal research wiki by hand means manually searching pages, cross-referencing claims, and losing the synthesized insight after each session. This Skill retrieves relevant wiki pages, synthesizes a cited answer, and optionally writes durable query results, concepts, or claims back into the knowledge base. ## Core Features & Use Cases - Grounded Q&A with citations: Reads context_brief.md, index.md, and up to 15 relevant pages, then synthesizes an answer where every key claim carries a [[slug]] wikilink to an existing page. - Crystallize mode: With --crystallize, writes the answer to wiki/outputs/, or creates new concept/claim pages, appends graph edges, and rebuilds context_brief.md and open_questions.md. - Gap awareness: Flags when a question touches known gaps in open_questions.md and suggests papers to ingest when the wiki lacks evidence. - Use Case: Ask "What is the core difference between LoRA and Adapter?" and receive a structured, cited comparison drawn from ingested paper pages, then crystallize it as a new cross-paper concept page. ## Quick Start Ask the wiki a research question such as "What is the core difference between LoRA and Adapter?" and optionally add --crystallize to save the synthesized answer back into the wiki.

Frequently Asked Questions about ask

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I ask a question against my research wiki?

Run the /ask skill from the wiki project root with a natural-language question, for example "/ask What is the core difference between LoRA and Adapter?". The skill reads context_brief.md and relevant pages, then returns a cited answer.

How do I save a synthesized answer back into the wiki?

Add the --crystallize flag to the ask command. The skill writes the answer to wiki/outputs/, or creates new concept or claim pages, appends graph edges, and rebuilds context_brief.md and open_questions.md.

What happens if the wiki has no relevant pages for my question?

The skill honestly reports that no relevant content exists instead of fabricating an answer. It suggests which papers to ingest and possible search directions such as arXiv keywords or Semantic Scholar queries.

Why does the ask skill fail with a missing context_brief.md error?

The skill requires wiki/graph/context_brief.md for global context. Rebuild it by running python3 tools/research_wiki.py rebuild-context-brief wiki/, then retry the question.

Can the ask skill answer questions beyond the wiki's content?

No. Answers must be grounded in actual wiki pages, and every citation must point to an existing page. When evidence is weak, the skill flags insufficient evidence and known gaps from open_questions.md.