Skip to content

Structure Wind-Tunnel Analysis

Finds the recurring underlying structure in a piece of information and stress-tests its causal logic

Stars
7,607
Licence
MIT
Files
2 files

What it does

Once installed, when you ask the agent to find the underlying structure of a problem, it looks past surface details for a recurring type of dilemma, draws the cause-and-effect relationships that make it work, and tests the idea by changing one condition to see where it breaks. It then gives a short conclusion with a transferable rule and a way to recognise it in the real situation. It can also save the result as a Markdown file in a local ~/Context folder.

Great for

  • People trying to understand why a meeting, team or project keeps repeating the same problem
  • Researchers or analysts who want a conceptual model to test, not just a summary
  • Anyone who wants more than a standard A/B test explanation of a messy situation

Skip it if

  • Simple fact lookups or summaries, where the skill adds unnecessary framing
  • Single-step execution tasks with one clear answer
  • High-stakes medical, legal or safety decisions: the skill says analogies can generate hypotheses but cannot replace domain evidence

Try asking

  • “找结构:为什么团队讨论总被第一个人带偏?”
  • “母题是什么?这家公司反复出现的部门冲突背后是什么机制?”
  • “用结构风洞分析:这个 AI 擅长深挖局部,却总错过全局异常。”
  • “不要只做AB测试,帮我看看这个增长停滞背后的结构。”
Step-by-step guide: how to use Structure Wind-Tunnel AnalysisInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 2 original files, the MIT licence and an attribution note.

    ljg-structure.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip ljg-structure.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip ljg-structure.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Give a concrete situation with actors and outcomes; vague prompts produce vague structures.
  • Ask for the analysis in read-only mode (say 只分析 or read-only) if you do not want a file written to ~/Context.
  • Treat the output as a testable hypothesis and check it against real observations, since the skill itself notes its examples do not prove the model is universally true.

What's inside

  • SKILL.md
  • Workflows/FindStructure.md

Redistributed unmodified from lijigang/ljg-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.