agentleFS
Balance0014Balance0014/claude-code-claim-check/skills/claim-check/SKILL.md

claim-check

Skill2 starsChanged 6 days agoMITView on GitHub

Use before telling the user that work is done, fixed, deployed or passing, or that something does not exist or is not used. Requires proof from this turn, or an honest "not verified".

11 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

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

  • 0

    Common

    Most of the 8 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:

    After your last code or config edit, run the thing that proves it: the tests, the build, the command, a request to the live URL.

    line 12
  2. Distinctive:

    Put the result in your answer: how many tests passed, the status code, the commit, the exit code.

    line 13
  3. Distinctive:

    If you edited again after that run, run it again.

    line 14
  4. Distinctive:

    If you can't run the check, say so: "changed, not verified yet".

    line 16
  5. Distinctive:

    Look at least twice, in different ways: search for the name, then search for a likely alias or read the place it would be registered.

    line 20
  6. Distinctive:

    Cleanup: No temporary files, debug output or backup copies left behind.

    line 30

“Common” is measured: kept by at least 6 in 10 of the 8 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

claim-check

Two kinds of sentence cause most of the damage when they are wrong: "it's fixed" and "it isn't there". Before you write either one, have the proof in this turn.

Before you say it's done, fixed, deployed or passing

  1. After your last code or config edit, run the thing that proves it: the tests, the build, the command, a request to the live URL.
  2. Put the result in your answer: how many tests passed, the status code, the commit, the exit code.
  3. If you edited again after that run, run it again. A check from before the last edit proves nothing about the code now.
  4. Re-reading the file you just edited is not a check.
  5. If you can't run the check, say so: "changed, not verified yet". That's an acceptable answer.

Before you say something doesn't exist or isn't used

Look at least twice, in different ways: search for the name, then search for a likely alias or read the place it would be registered. If you only looked once, call it a guess.

When "done" means a real fix

A real fix covers five things. Say which ones you did, and which you skipped:

  1. Cause. You found why it broke, not only where.
  2. Prevention. Something now stops the same failure, such as a test, a check or a guard.
  3. Same bug elsewhere. You looked for the same bug in similar places.
  4. Verification. You ran the real thing and saw it work.
  5. Cleanup. No temporary files, debug output or backup copies left behind.

A workaround that stops the symptom is fine. Call it a workaround, not a fix.

If a hook blocked you

The claim-check hook found a claim without proof. Run the check and show the result, cite proof you already have (a commit hash, a line like "14 passed" or "HTTP 200") if you changed no code this turn, or rewrite the sentence as "not verified yet". Do one of these before you stop again.

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.