agentleFS
smarzbansmarzban/agent-skills/skills/writing-readmes/SKILL.md

writing-readmes

Skill1 starsChanged 3 days agoMITView on GitHub

Write, overhaul or review a repo's front-door README.md. The first screen says what the project is and who it is for, the quickstart is proven by a cold run, and reference material moves to docs. Use for "write the README", "fix the README", "review the README", "the readme isn't clear", or a README that has grown into a docs dump.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. smarzban/agent-sdlc★ 2same 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.

  • In this repository15 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every other command, flag, default and path: open the code that defines it.

    line 33
  2. Very specific:

    If the repo is public, check that its description and topics say the same thing as the first sentence.

    line 43

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

writing-readmes

A README is the front door. On the first screen a newcomer learns what the project is, whether it is for them, and how to start. Everything deeper lives in the docs and is linked.

Work on the current repo unless you are given a path. In a monorepo, "the README" is the root README; write a package README only when asked for it. Follow reference/style.md for the prose.

In Write mode, when you are in the repo's main checkout (git rev-parse --path-format=absolute --git-dir --git-common-dir prints the same path twice), do not edit there: another session may be working in it. Do the whole run on a new branch in a worktree of your own (from the repo root, git worktree add -b readme ../<repo>-readme; if that branch or folder exists, pick another name, never reuse or delete it) and give its path and branch in the report. Work in place when the repo has no commits or the README has uncommitted changes.

Modes

  • Write: produce or overhaul the README. This is the default.
  • Review: report only, change nothing. Use it when asked to review, check or audit a README. Run steps 1 and 4 against the existing file and list each finding with its line and the evidence.

Steps

<skill> in a command is this skill's folder, the one that holds this file. Run those scripts by that path, with the target repo as the working directory.

  1. Read the repo before the README: the package manifest, the entry points, the CI config, the docs folder, the licence file. Settle the repo type (library, CLI, service, application, framework, monorepo, skills bundle) and the reader. Then read the existing README and mark each claim true, stale or unverifiable against the code.
  2. Choose the sections from reference/anatomy.md for that repo type. Leave out a section that has nothing true to say.
  3. Write.
    • First screen: the name, one sentence that says what it is and who it is for, then three to six highlights. Each highlight is a fact a reader can check (a command, a supported platform, a limit, a measured number), not a claim about quality.
    • Quickstart: the shortest path from nothing to a visible result, in copyable blocks with the expected output. Keep the full install (other platforms, options, configuration) in its own section or in the docs.
    • Move reference material (the full configuration table, every command, deployment) to the docs and leave a two-line summary with a link. Move it; do not delete it. If no docs page exists for it, create one or keep the material and report it.
    • Contributing: link to AGENTS.md and CONTRIBUTING.md for the commands and rules. Do not repeat them.
    • For a project that agents consume (a CLI, an MCP server, a library, a skills bundle), add the Use with agents section from reference/anatomy.md.
  4. Verify.
    • Quickstart: give a fresh subagent only the README and a clean checkout (git clone into a temporary folder, or git worktree add), and tell it to follow the quickstart exactly and report each command, its output, and where it had to guess. A guess is a defect in the README. Tell it also to stop before a step that would change what this machine already has in use (a global install, moving an image tag that exists, starting, stopping or reconfiguring a service the quickstart does not own, such as Docker) and to report that step instead; report it as not proven, and ask the owner before running it. Without a subagent, run the quickstart yourself in a clean checkout under the same limit and say that the cold run was not independent.
    • Every other command, flag, default and path: open the code that defines it.
    • Links and paths: run <skill>/scripts/check-links README.md. It must report 0 broken.
    • Badges: each badge points at something that exists. The workflow file named in a CI badge is in .github/workflows/, the package named in a version badge is published, and the licence badge matches the LICENSE file. Remove a badge that fails this.
    • Images: every image path resolves. Never link an image that is not in the repo; list it as something for the owner to supply.
  5. Report: what the README now covers, what moved to which docs page, what you could not verify, and the owner decisions.

Owner decisions

  • Licence. For a public repo, or one about to become public, a missing LICENSE file is reported, and the README gets no licence section until the owner picks one. Never choose a licence. For a private repo, do not raise it.
  • Screenshot or demo. Propose where one would help. Do not fabricate one.
  • If the repo is public, check that its description and topics say the same thing as the first sentence. publishing-a-repo covers discovery in full.
  • llms.txt belongs to a hosted docs site, not to a README. Do not add one to a repo without a site.

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.