agentleFS
Sign inStart free
iamzhanggiamzhangg/skill-benchmark/SKILL.md

skill-benchmark

Skill0 starsChanged 49 days agoMITView on GitHub

Design and run lightweight behavioral evaluations for Agent Skills using realistic scenarios, observable assertions, boundary cases, and regression reports. Use when proving a Skill works or comparing revisions; not for judging prose style from a single cherry-picked example.

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 3

    1. Rifteo/skills★ 37same name
    2. AllanSantos-DV/skill-kit★ 0same name
    3. TheophilusChinomona/idev★ 0same 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.

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 5

    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:

    Keep fixtures free of secrets and network dependencies.

    line 15
  2. Distinctive:

    Run each scenario in a clean workspace with the target Skill and save the final transcript or artifact manifest.

    line 19
  3. Distinctive:

    Do not silently rewrite failing expectations to match the latest output.

    line 25
  4. Distinctive:

    Treat safety and authorization failures as blocking even if the aggregate score is high.

    line 30
  5. Distinctive:

    Use references/reporting.md for the final summary and regression decision.

    line 34

“Common” is measured: kept by at least 6 in 10 of the 3 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

Skill benchmark

Turn a Skill's promises into repeatable evidence. Prefer a few discriminating scenarios over many shallow examples.

Design

  1. Extract capabilities and boundaries from the Skill description and body.
  2. Define a balanced suite: happy path, ambiguous input, missing prerequisite, unsafe or out-of-scope request, and regression case when applicable.
  3. Make assertions observable. Supported deterministic assertions are documented in references/case-format.md.
  4. Keep fixtures free of secrets and network dependencies. Separate the prompt/fixture from the expected behavior so test data does not leak the answer.

Execute

Run each scenario in a clean workspace with the target Skill and save the final transcript or artifact manifest. Then grade it:

python scripts/grade.py cases.json results.json --output report.md

Do not silently rewrite failing expectations to match the latest output. Diagnose whether the failure belongs to the Skill, evaluator, fixture, or environment.

Interpret

  • Report assertion-level results, not just a single score.
  • Treat safety and authorization failures as blocking even if the aggregate score is high.
  • Distinguish deterministic checks from model-judged rubrics. If using an LLM judge, record model, prompt, temperature, and repeated-run variance.
  • A benchmark demonstrates behavior under its scenarios; it does not prove universal reliability.

Use references/reporting.md for the final summary and regression decision.

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