agentleFS
aniq-ui-proaniq-ui-pro/agent-skills/skills/docs-check/SKILL.md

docs-check

Skill1 starsChanged 2 days agoMITView on GitHub

Proves a setup guide (README, QUICKSTART, INSTALL) can be followed word for word on a clean copy of the project. Use after writing or changing setup docs, before a release, or when someone says the README does not work. A script checks every command, file, env variable, port and local URL the guide names; then you follow the guide yourself and fix every place you had to guess.

  • Reads credentials
  • Deletes or force-pushes

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 0

    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. Very specific:

    Check: every command exits 0 (echo $?).

    line 46
  2. Very specific:

    Stop only what you started (the PIDs you launched, docker compose down in the copy), then delete the copy.

    line 73

“Common” is measured: kept by at least 6 in 10 of the 5 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

docs-check

A guide passes when a new reader can type its commands in order, on a fresh copy, and the last step shows the app answering, with no guessing. Work on a copy, never in the folder you develop in: your node_modules, .env and build output hide every gap.

1. Pick the guide and what the reader gets

Name the first file a new user opens (usually README.md) and the folder they get: a fresh clone, or the ZIP or package you ship. Check: you can write down both paths.

2. Make a clean copy

rm -rf /tmp/docs-check-copy && mkdir /tmp/docs-check-copy
git archive HEAD | tar -x -C /tmp/docs-check-copy   # or unzip the release there

Check: ls -a /tmp/docs-check-copy shows no node_modules, .next, dist or .env.

3. Run the static check

cd /tmp/docs-check-copy
node ${CLAUDE_SKILL_DIR}/scripts/docs-check.mjs README.md QUICKSTART.md

It only reads files. Add --root . when a guide in a subfolder talks about its siblings, --ignore <path|VAR|port> for something your packaging adds, --all to list every pass. For each FAIL, read the guide line and the project file yourself, then fix the guide, fix the project, or (if the script cannot see it, like a folder the app creates at runtime) ignore it and write down why. Read every WARN the same way. Check: the last line says 0 fail, and every ignore has a reason.

4. Follow the guide word for word

Open a new terminal where the guide says to. Type each command exactly as written, in order, adding nothing. Skip deploy steps and anything that drops data or touches a live server. Log each command with its exit code and how long it took. Check: every command exits 0 (echo $?). A non-zero exit is a finding: stop there.

5. Write down every guess

Each time you do something the guide did not say (which folder or terminal, a tool to install, a value to copy, how long to wait, a password), add a line to GUESSES.md: the guide line, what you did, what the guide should say. Check: GUESSES.md exists, even if it says "none".

6. Prove the last step

The guide must end with the app answering. For every address it gives:

curl -sS -o /dev/null -w '%{http_code}\n' http://localhost:3030/

Check: each address returns 200 (or what the guide says it returns), and the page opens in a browser.

7. Fix the guide, then run it again

Fix the real repository, not the copy. Commit, make a new copy, and repeat steps 3 to 6. Check: step 3 ends with 0 fail and GUESSES.md says "none".

8. Clean up

Stop only what you started (the PIDs you launched, docker compose down in the copy), then delete the copy. Check: docker ps and lsof -i :<port> show nothing you started.

What the script cannot see

Routes registered by libraries or rewrites, folders the app creates when it runs, external links, and whether a command succeeds. Steps 4 to 6 cover those.

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.