ljg-qa
lijigang/ljg-skills
Extract core arguments as incisive Q-A chains that reconstruct an author's reasoning, not summaries or study aids.
What is ljg-qa?
ljg-qa transforms articles, papers, or books into structured question-answer pairs that expose the author's logical skeleton. Each question cuts to the core (why, trade-offs, boundaries), and each answer is formally closed with conclusion, formalized relationship, reasoning steps, and limits. Use when you need intellectual scaffolding—the reader walks the Q chain and nails down each insight, not just receives conclusions.
- Extracts argument structure as a directed Q chain (not parallel bullet points)
- Formulates questions that demand substantive answers—why a solution works, where it fails, what it costs—not definitional "what is X" questions
- Structures each answer in four parts: one-line conclusion, formalized relationship (using text + symbols like → = ≠), reasoning steps, and boundary conditions
- Outputs in org-mode format to ~/Documents/notes/ with denote-style filenames
- Handles URLs, PDFs, and direct text input without requiring external setup
How to install ljg-qa
npx skills add https://github.com/lijigang/ljg-skills --skill ljg-qaHow to use ljg-qa
- 1.Invoke with /ljg-qa followed by a URL, file path (PDF), or paste text directly
- 2.The skill will fetch content (if URL/file), identify the argument structure, and design a Q chain based on logical dependencies
- 3.Each question is formulated to expose core tensions, trade-offs, or boundaries in the author's reasoning
- 4.Answers are written in four strict parts: conclusion, formalized relationship, reasoning, and limits
- 5.Output is saved as an org-mode file in ~/Documents/notes/ with a denote-style timestamp and topic slug
Use cases
- Extract the core logic from a research paper to understand its method's trade-offs and failure modes
- Convert a long article into a Q-A scaffold that guides readers through the author's reasoning path
- Decompose a book chapter to expose the underlying argument structure and dependencies between ideas
- Identify the "why" and "when not" behind a technical approach by walking its Q chain
- Researchers and students who need to understand arguments deeply, not just summarize them
- Knowledge workers building mental models from complex texts
- Anyone extracting ideas for teaching or knowledge management who wants scaffolding, not study quizzes
ljg-qa FAQ
A summary compresses content; an FAQ answers isolated questions. ljg-qa reconstructs the author's reasoning chain—each Q flows from the previous A, and each A is structured to expose logic, not just convey facts. It's intellectual scaffolding, not study aids.
A good Q cuts to the core: why does this solution work, what are its costs, where does it fail, how does it differ from alternatives? Bad Qs can be answered in one sentence ("what is X"). Good Qs demand substantive answers that carry weight.
Conclusion anchors the idea; formalized relationship (using text + symbols) makes the logic visible at a glance; reasoning steps show how to arrive there; boundary conditions prevent overgeneralization. Together, they let readers replicate the author's thinking.
Best for argumentative or explanatory content: research papers, essays, technical articles, book chapters. Less useful for narrative, fiction, or purely factual lists where the goal is not to expose a reasoning chain.
It's visual shorthand using text and simple symbols (→, =, ≠, +, ×) to compress a relationship into one readable line. Example: `通才 = 协调，专才 = 干活` (generalist = coordination, specialist = execution). It's the geometry of thought, not math.
Full instructions (SKILL.md)
Source of truth, from lijigang/ljg-skills.
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. user_invocable: true
ljg-qa: 问答提取
读一份东西，把它的思想拆成「为什么—怎么—边界」的问答链。
读者顺着 Q 走过去，每个 A 砸下来一枚钉子。
你不是
- 不是 FAQ 生成器（"什么是 X"——读者一看就跳过）
- 不是摘要换皮（把段落拆成"问/答"两半还是摘要）
- 不是知识点列表（孤立的事实碰撞不出洞察）
- 不是阅读理解题（提问不是为了考读者，是为了切中作者）
你是
把作者的论证骨架翻出来，每根骨头长成一个尖锐的问题。读者沿着 Q 链读，能复现作者的整套思路——而不是被告知结论。
三条铁律
-
Q 切要害 —— 问的是「为什么这个解法成立」「它跟另一种做法差在哪」「它的代价是什么」「它在哪里失效」，不是「它定义是什么」。一个 Q 必须能让答案承重，不能被一句话敷衍过去。
-
A 有形式化收口 —— 每个 A 严格四段：结论（一句话）+ 形式化（用文字 + 简单符号把思想压成一行可视关系，如
A = B + C、旧: X → 新: Y）+ 论证步（怎么想到的）+ 边界（不成立的条件）。形式化是"思想的几何"，让读者一眼看出关系。 -
Q 链有方向 —— Q 之间不是并列罗列，是「Q1 答完→Q2 自然冒出来」。读者读完整串 Q，相当于走了一遍作者的推理路径。
工作流
按 Workflows/Extract.md 的步骤执行。
设计参考
Q 怎么提、A 怎么收口的具体模式见 References/QuestionDesign.md。
Voice Notification
执行 workflow 时：
curl -s -X POST http://localhost:31337/notify \
-H "Content-Type: application/json" \
-d '{"message": "Running Extract in ljg-qa"}' \
> /dev/null 2>&1 &
输出文本：
Running **Extract** in **ljg-qa**...
输出
- 格式：org-mode（
*bold*，禁 markdown 语法） - 路径：
~/Documents/notes/ - denote 文件名：
{YYYYMMDDTHHMMSS}--qa-{核心主题 5-10 字}__qa.org
Examples
Example 1: URL
User: /ljg-qa https://example.com/article
→ WebFetch 获取
→ 找观点骨架 → 设计 Q 链 → 写 A 三段
→ org-mode 输出到 ~/Downloads/
Example 2: 论文 PDF
User: /ljg-qa ~/Downloads/paper.pdf
→ Read PDF（注意 pages 参数）
→ Q 抽出方法的「为什么」「代价」「边界」
→ 输出 org-mode
Example 3: 直接文本
User: 把这段抽成 Q-A: [text]
→ 跳过获取，直接抽
→ 输出
Gotchas
- AI 默认会写「什么是 X」型问题 —— 教科书腔。生成后扫一遍，凡是 Q 能用一句定义打发的，重写
- AI 默认会让 A 散掉 —— 没有结论句、没有边界、写成一段散文。每个 A 必须严格四段（结论 / 形式化 / 步骤 / 边界）
- AI 默认会把「形式化」写成数学公式 —— 不是。形式化是用文字 + → = ≠ + × 这类符号压一行可视的关系，比如
通才 = 协调，专才 = 干活。是"思想的几何"，不是"数学的形式" - AI 默认按章节顺序提问 —— 这是抄目录，不是抽思想。Q 链应该按论证依赖关系排，不按出现顺序
- AI 默认会把 Q-A 理解成「问答游戏」 —— 不是。这里 Q 是凿子，A 是钉子。装饰性的轻问题禁止
- AI 默认会在 A 里堆术语保平安 —— 用术语不算回答。把术语翻译成具体动作和具体物件，否则 A 没承重
Related skills
More from lijigang/ljg-skills and the wider catalog.
ljg-travel
Deep cultural travel research: auto-generate structured org-mode documents and reference cards for museums and ancient architecture.
ljg-card
Transform content into styled PNG visuals with seven layout molds—long cards, infographics, comics, sketchnotes, whiteboards, and more.
ljg-roundtable
Structured roundtable discussion framework with truth-seeking moderator and representative figures for multi-perspective debate.
ljg-paper
Reads research papers as continuous stories, extracting core propositions through narrative structure for non-academics.
ljg-plain
Rewrite any content so a smart 12-year-old understands it—plain language, no jargon.
ljg-learn
Deep concept anatomist that deconstructs ideas through 8 exploration dimensions and compresses insights into epiphany.