AI Skills
ljg-qa

ljg-qa

Popular

Extracts core ideas from an article, paper, or book as a chain of incisive Q-A pairs. Questions cut to the chase, not textbook-style; answers are concise, formally closed, with a complete logical chain. Readers follow the Q chain, each answer driving a nail, reproducing the author's full reasoning. Triggers on user input like '问答', 'Q&A', 'QA', '提问', '抽取问题', '/ljg-qa', or sharing content for Q-A extraction. NOT for FAQ generation, glossary creation, or comprehension quizzes — this is intellectual scaffolding, not study aids.

6.6Kstars
767forks
Updated 7/17/2026
Get SkillSource Code

Related Skills

writing-skills233Kdoc-coauthoring153Kbrainstorming45K
SKILL.md
readonlyread-only
name
ljg-qa
description

信息提问机。给一篇文章/论文/书，把核心观点抽成 Q-A 对——Question 切要害，不教科书；Answer 简洁清晰，有形式化收口，逻辑链完整。读者顺 Q 链走过，每个 A 砸下一枚钉子，复现作者整套推理。Use when user says '问答', 'Q&A', 'QA', '提问', '抽取问题', '/ljg-qa', or shares an article/paper/book and asks for Q-A extraction. Triggers when the user wants ideas extracted not as a summary but as a sequence of incisive questions with answered. NOT FOR FAQ generation, glossary creation, or comprehension quizzes — this is intellectual scaffolding, not study aids.

ljg-qa: Q-A Extraction

Read a piece of content and break its ideas into a chain of "why—how—boundary" Q-A pairs.

Readers walk through the Q chain, each answer driving a nail.

You Are Not

  • Not a FAQ generator ("What is X" — readers skip it)
  • Not a summary in disguise (splitting paragraphs into "question/answer" halves is still a summary)
  • Not a list of knowledge points (isolated facts don't spark insight)
  • Not reading comprehension questions (questions aren't to test the reader, but to cut to the author)

You Are

Expose the author's argument skeleton, each bone shaped into a sharp question. Readers following the Q chain can reproduce the author's full reasoning — not just be told the conclusion.

Three Hard Rules

  1. Q cuts to the chase — Ask "why does this solution work", "how is it different from another approach", "what is its cost", "where does it fail", not "what is its definition". A Q must make the answer bear weight; it can't be brushed off with one sentence.

  2. A has formal closure — Each A strictly four parts: Conclusion (one sentence) + Formalization (use text + simple symbols to compress the idea into a visible relationship, e.g., A = B + C, 旧: X → 新: Y) + Reasoning steps (how we got there) + Boundary (conditions where it doesn't hold). Formalization is "geometry of thought" — lets readers see relationships at a glance.

  3. Q chain has direction — Qs are not a parallel list; they follow "after Q1 is answered, Q2 naturally emerges". Reading the full chain is like walking the author's reasoning path.

Workflow

Follow steps in Workflows/Extract.md.

Design Reference

Specific patterns for how to ask Qs and close As are in References/QuestionDesign.md.

Voice Notification

When running the workflow:

curl -s -X POST http://localhost:31337/notify \
  -H "Content-Type: application/json" \
  -d '{"message": "Running Extract in ljg-qa"}' \
  > /dev/null 2>&1 &

Output text:

Running **Extract** in **ljg-qa**...

Output

  • Format: org-mode (*bold*, no markdown syntax)
  • Path: ~/Documents/notes/
  • Denote filename: {YYYYMMDDTHHMMSS}--qa-{核心主题 5-10 字}__qa.org

Examples

Example 1: URL

User: /ljg-qa https://example.com/article
→ WebFetch fetches
→ Find argument skeleton → Design Q chain → Write A three parts
→ org-mode output to ~/Downloads/

Example 2: Paper PDF

User: /ljg-qa ~/Downloads/paper.pdf
→ Read PDF (note pages parameter)
→ Extract Qs for method's "why", "cost", "boundary"
→ Output org-mode

Example 3: Direct text

User: Extract this into Q-A: [text]
→ Skip fetch, extract directly
→ Output

Gotchas

  • AI defaults to writing "What is X" questions — textbook tone. After generation, scan: if a Q can be answered with a definition, rewrite it.
  • AI defaults to letting A become loose — no conclusion sentence, no boundary, written as prose. Each A must strictly have four parts (conclusion / formalization / steps / boundary).
  • AI defaults to writing formalization as math formulas — no. Formalization uses text + → = ≠ + × symbols to compress a visible relationship, e.g., 通才 = 协调，专才 = 干活. It's "geometry of thought", not "mathematical form".
  • AI defaults to asking questions in chapter order — that's copying the table of contents, not extracting ideas. The Q chain should follow argument dependency order, not appearance order.
  • AI defaults to treating Q-A as a quiz game — no. Here Q is a chisel, A is a nail. Decorative lightweight questions are forbidden.
  • AI defaults to piling on jargon for safety — using jargon is not an answer. Translate jargon into concrete actions and objects, otherwise A has no weight.