agentleFS
chinmay-sawantchinmay-sawant/gopdfsuit/skills/feynman/SKILL.md

feynman

Skill607 starsChanged 59 days agoMITView on GitHub

Feynman technique. Explain a topic in plain words, hunt your own gaps, fill them from source, repeat until clean. If you cannot explain it simply, you do not understand it yet. Run with /feynman <topic>.

We split this skill into the rules it gives an agent

6 rules. 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before any response, run the /unslop skill on your output and apply its fixes.

    line 11
  2. Distinctive:

    No topic? Ask once, wait for the answer.

    line 21
  3. Distinctive:

    Never explain from memory alone.

    line 25
  4. Distinctive:

    If sources contradict what you believed, trust the sources and say what changed.

    line 29
  5. Distinctive:

    Every step names a mechanism, not a reputation.

    line 46
  6. Distinctive:

    After flags, drill one question at a time: "why does that work?" until they hit bedrock or hit a wall.

    line 57

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

The rule: if you cannot explain something in a simple way, you do not fully understand it yet. This skill runs that rule as a loop until the explanation holds. In gopdfsuit use it for PDF engine ideas such as font subsetting, xref tables, compress tiers, XFDF fill, tagging and ParentTree, or Sonic pooling.

Unslop first

Before any response, run the /unslop skill on your output and apply its fixes. Simple words are the whole point here; an explanation padded with "leverages" and "robust frameworks" fails its own test. Unslop, then send.

Invocation

Parse the topic from the invocation. Three forms:

  • /feynman <topic>: you explain the topic.
  • /feynman this: explain whatever the conversation was just about.
  • /feynman me <topic>: role flip. The user explains, you listen and grade.

No topic? Ask once, wait for the answer.

Grounding

Never explain from memory alone.

  • Topic touches this repo: read the actual code first. Every claim about it cites file:line. For gopdfsuit start in internal/pdf/, pkg/gopdflib/, internal/handlers/, or internal/models/ and name the exact file.
  • Topic outside the repo: read authoritative docs or source before explaining, and say what you read.
  • If sources contradict what you believed, trust the sources and say what changed.

The loop

One pass has three moves. Repeat until the exit gate passes.

  1. Explain it to a smart 12-year-old who shares none of your vocabulary. Short sentences. Concrete analogies. Every technical term either avoided or defined in the same breath it first appears.
  2. Audit your own explanation for gaps. Label each finding:
    • Jargon: a term the 12-year-old would not know, used without definition.
    • Hand-wave: "it just handles that", "under the hood". No mechanism named.
    • Circularity: X explained by Y, Y explained by X.
    • Name-dropping: leaning on a famous name instead of saying what it does.
  3. Fill each gap at the source: read the code or doc covering exactly that fuzzy spot. Then re-explain, tighter than last time.

Exit gate (all must hold):

  • Zero audit findings on the latest pass. Not fewer than last time. Zero.
  • Every step names a mechanism, not a reputation.
  • The user could retell the explanation correctly without you in the room.

Report the pass count. "Clean in 3 passes" means something; "clean" alone does not.

Grading mode (/feynman me <topic>)

You become the listener.

  • Let the user finish a full attempt before interrupting.
  • Flag findings with the same four labels above, quoting their exact words back.
  • After flags, drill one question at a time: "why does that work?" until they hit bedrock or hit a wall. A wall is the method working, never mock it.
  • Point every gap to the file or doc that closes it. For gopdfsuit point to internal/pdf/*, documentation/TEMPLATE_REFERENCE.md, or sampledata/<area>/ fixtures.
  • End by applying the exit gate to their final version and naming what still would not survive retelling.

Download now (ZIP)

  • Other versions10 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 1

    1. chinmay-sawant/developer-skills★ 097% same

    Wrote their own version · 9

    1. LilMGenius/paperthin★ 1.1ksame name
    2. liza-mas/liza★ 414same name
    3. fainshare/Liam-Skills★ 111same name
    4. neurofoo/agent-skills★ 111same name
    5. oakreader/oakreader★ 46same name
    6. satsilem/claude-skills-pantheon★ 4same name
    7. nuttakulsv/think-like★ 2same name
    8. AlexHorovitz/skills★ 1same name
    9. vansvost-arch/claude-skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • Used alongsideProjects with it also use socrates.

    Other files people use alongside this one.

    Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
    • zhengminjie1981Skill

      socrates

      苏格拉底式思考伙伴,通过提问引导用户发现答案。

      Also in 3 projects that use this

  • In this repository4 other files it gives its agents.

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.