agentleFS
Sign inStart free
elon-chooelon-choo/fablever/.agents/skills/fable-evidence-done/SKILL.md

fable-evidence-done

Skill33 starsChanged 5 months agoMITView on GitHub

Before you say "done", "fixed", "works", "passing", or "verified", ground that claim in a real command/test/file/tool result on the same line — or say plainly that it is not verified yet. Use whenever you are about to report completion or success of a code change, a build, a fix, or a task. Catches unsupported "it works" claims. Pulled on demand; not always-on.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    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 completion claim names its evidence on the same line, or it is downgraded to "not verified".

    line 26
  2. Distinctive:

    Before claiming completion, find the evidence: Run the test, check the exit code, read the file, diff against the spec, hit the endpoint — …

    line 35
  3. Very specific:

    Cite it inline: Put the command/result next to the claim ("done — <cmd> → <result>").

    line 37

≈ “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

fable-evidence-done — never claim it works without evidence it works

Decisiveness has an honest cost. The style-only ablation showed the Fable working style raised unsupported "it works" claims to 8.3% vs plain Claude's 2.1% (eval/style-only-ablation/) — the trade for being outcome-first. fablever mitigates this in the tool with fable_lint's unsupported-done-claim rule (label regression: 100% on the fixture set, eval/unsupported-claim-regression/). This skill is the behavior that rule guards: a completion claim must carry its evidence.

When to use this

  • You are about to write "done / fixed / works / passing / verified / resolved / ready" (or the same in another language) about a code change, build, fix, migration, or task.

When NOT to use this

  • You are describing a plan or an intention, not asserting a result.
  • A pure-conversation answer with no executed work to verify.

The rule

Every completion claim names its evidence on the same line, or it is downgraded to "not verified".

  • "Fixed. It works now." → not allowed (no evidence).
  • "Fixed — npm test passes (42/42)." → allowed (names the check).
  • "Implemented, but not verified yet — haven't run the suite." → allowed (honest, no false claim).
  • "고쳤고 작동합니다." → not allowed. "고쳤습니다. npm test로 확인했습니다." → allowed.

Procedure

  1. Before claiming completion, find the evidence. Run the test, check the exit code, read the file, diff against the spec, hit the endpoint — a check that could actually fail.
  2. Cite it inline. Put the command/result next to the claim ("done — <cmd> → <result>").
  3. If you have not verified, say so. "Implemented but not verified" is a complete, honest report; a bare "it works" is not. Do not run the verification only in your head.
  4. Skip the ceremony on trivial one-liners — a typo fix does not need a test harness, but also does not need a triumphant "works!".

Expected output

Completion claims that each carry an inline, checkable piece of evidence — or an explicit "not verified".

Verification

Re-scan your final message for "done/works/fixed/verified". For each, is there a command/test/file/tool result on that line? If not, either run the check now or downgrade the wording.

Failure cases

  • "All green, ship it" with no test actually run (tone-only completion claim).
  • Saying "verified" when the verification happened only in reasoning, not as a tool result.
  • Reporting a passing test you did not run this session.

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.

Check your setup, free