agentleFS
robinbrilrobinbril/claude-code-harness/skills/readme/SKILL.md

readme

Skill3 starsChanged 23 days agoMITView on GitHub

Layout, wording and visuals for a README and the one-line GitHub repo description - problem first, sharp expert copy, custom light/dark SVG images, verified in GitHub's own renderer. Use for /readme, a new README, a README overhaul or a repo description.

32 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use favicon, build and 5 more.

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.
  • cpiprintSkill

    favicon

    Generate favicons from a source image

    Also in 3 projects that use this

  • Microsoft AzureSkill

    build

    Build the TypeSpec Rust emitter from source. Run this after cloning the repo, after pulling changes, or when TypeScript source files have been modified.

    Also in 3 projects that use this

  • GitHubSkill

    refactor

    Surgical code refactoring to improve maintainability without changing behavior. Covers extracting functions, renaming variables, breaking down god functions, improving type safety, eliminating code smells, and applying design patterns. Less drastic than repo-rebuilder; use for gradual improvements.

    Also in 3 projects that use this

  • BlockSkill

    plan

    Draft a TDD, PRD, or lightweight plan in the conversation without committing it.

    Also in 3 projects that use this

  • SentrySkill

    commit

    Create commit messages following Sentry conventions. Use when committing code changes, writing commit messages, or formatting git history. Follows conventional commits with Sentry-specific issue references.

    Also in 3 projects that use this

  • NVIDIASkill

    research

    Before designing a feature, research how comparable libraries solve the problem, check best practices, known pitfalls, and upstream changes. Use when starting a new feature, investigating an unfamiliar area, or the user asks to research something.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 32 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read manifests, entry points, install scripts, existing docs and the license files.

    line 12
  2. Distinctive:

    Use this order. Drop a section only when it has nothing true to say.

    line 16
  3. Distinctive:

    Tradeoff: when not to use it, honestly.

    line 26
  4. Distinctive:

    Sources and licenses: every borrowed part with link and license.

    line 27
  5. Distinctive:

    Write for a sharp, busy engineer.

    line 33
  6. Very specific:

    Commit the spec next to the SVGs (assets/visuals.json), so anyone can change a word and regenerate.

    line 57
  7. Very specific:

    After pushing, open the README on GitHub itself and confirm every assets/ image has loaded.

    line 73
  8. Distinctive:

    Every claim, count, link and command checked against the repo.

    line 77

“Common” is measured: kept by at least 6 in 10 of the 32 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

README

A README has one job: a stranger understands in ten seconds what this is, why they want it and how to start. It reads like a serious engineering product: confident, precise, impressive. Facts come from the repo, visuals from a spec you can regenerate.

1. Collect facts first

Read manifests, entry points, install scripts, existing docs and the license files. Count what you will claim (skills, commands, checks) from the source, never from memory. Every link, command and version in the README must exist.

2. Tell it as a story

Use this order. Drop a section only when it has nothing true to say.

  1. Banner with name, one-line promise and the 3 to 5 parts; under the badges a bold one-line positioning statement.
  2. The problem: the failure modes the reader already suffers, ideally as a table of failure, what happens, what this project does about it.
  3. What you get: one table, layer and what it enforces or delivers.
  4. How it works: one flow visual, four to six steps.
  5. Install: the shortest route first, one command per block; the alternative after it.
  6. Featured: the one thing people come for, with "you type / you get" examples.
  7. Reference: tables per group (what, when). Names the reader types stay in backticks.
  8. How you know it works: observable signs, not promises.
  9. Tradeoff: when not to use it, honestly.
  10. Sources and licenses: every borrowed part with link and license.

The problem, "how you know it works" and "tradeoff" sections follow the README of andrej-karpathy-skills: naming the failure first is what makes a reader recognise themselves.

3. Sharp, not simple

  • Write for a sharp, busy engineer. Clear is not childish: never explain the obvious, never use toddler wording ("klikproeven", "bewakers") for a term the audience knows (e2e tests, hooks).
  • Precise domain terms are the impressive part: quality gate, root cause, deterministic, scope drift. Explain a term once only when the audience may not know it.
  • Lead each line with the strongest concrete fact: a mechanism, a guarantee, a number. Impressive comes from specifics, not adjectives.
  • Short sentences, active verbs. No hype words ("powerful", "revolutionary", "seamless"), no em-dashes, no emoji unless the project uses them.
  • Table columns answer "what" and "when", never restate the name.
  • Numbers that can drift (counts, versions) live in text and badges, never inside images.

4. Repo description

The one line under the repo name on GitHub (About) and in package or plugin metadata.

  • One sentence, at most about 140 characters: who it is for or what it becomes, then what is in it.
  • Same promise as the banner tagline, in the same language as the README.
  • Concrete nouns over adjectives; no "powerful", "simple" or "AI-powered".
  • Changing the About text is public: show the old and new text and get a yes first, then gh repo edit <eigenaar/repo> --description "<tekst>" and read it back.

5. Visuals

Make banner and flows with scripts/make_svgs.py from a JSON spec; see references/spec-voorbeeld.json for the shape and ICONS in the script for icon names.

python3 <deze-skill-map>/scripts/make_svgs.py assets/visuals.json assets
  • Commit the spec next to the SVGs (assets/visuals.json), so anyone can change a word and regenerate.
  • Each SVG carries its own light and dark colours; no Mermaid for the main visuals.
  • The script stops when a text will not fit. Shorten the text, do not shrink the font.
  • Embed with <img src="assets/x.svg" alt="what the image says" width="100%">.
  • Badges: shields.io for version, counts and license; the real CI badge when CI exists.

6. Verify

  1. Render with GitHub's own markdown engine, light and dark, at 1012 and 390 px:

    node <deze-skill-map>/scripts/render_readme.mjs README.md <out-map> <eigenaar/repo>

    The report must say every image loads and nothing scrolls sideways.

  2. Read the screenshots. Check the banner, the first table and one flow in both themes.
  3. After pushing, open the README on GitHub itself and confirm every assets/ image has loaded.

Done when

  • Every claim, count, link and command checked against the repo.
  • Screenshots read in light and dark mode; the report is clean.
  • The repo description matches the README's promise.
  • A first-time reader can answer: what is it, why would I want it, how do I start.

Download now (ZIP)

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.