Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →hyhmrright/new-skill
Scaffold a new brooks-lint analysis skill so it passes `npm run validate` and `npm run evals` on the first try — generates skills/{name}/SKILL.md (with the mandatory "Do NOT trigger for:" clause and a Process section citing guide step ranges) plus skills/{name}/{name}-guide.md (sequentially numbered steps), then appends paired eval scenarios.
Passed 16 of 21 test cases with the skill, and 3 of 21 without it.
and 12 more in the full report
15 fixed3 passed either way6 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
This benchmark measured an earlier version of the skill.
A newer attempt on gemini-3.8-flash on 10/19/2026 did not produce a publishable comparison (not comparable). The snapshot above is the previously measured run.
One of 15 test cases the skill fixed. Same model, same prompt, with and without the skill.
After finishing local testing of a skill scaffold, what slash commands restore the official marketplace plugin copy?
| Checked | Without skill | With skill |
|---|---|---|
| Executes /plugin marketplace update followed by /plugin install | Missed | Passed |
The instructions your model reads when the skill loads.
Skill name comes from the user request (kebab-case, e.g. brooks-security). If empty,
ask for the name and a one-line purpose first.
Follow the repo's "Adding a New Skill" checklist exactly:
skills/{name}/SKILL.md with frontmatter — name, a description
that ends with a Do NOT trigger for: clause (validate-enforced for shipped
skills; omitting it causes false triggering), metadata: with opencode/slash: "true", and a Process section of 3–6
bullets that cite the guide's step ranges inline (e.g. Scan risks (Steps 1–6 of the guide)). Mirror the structure of an existing skill such as
skills/brooks-review/SKILL.md (Setup → Process → Mode line).skills/{name}/{name}-guide.md with sequentially numbered steps —
no gaps, no duplicates. Sub-steps like Step 2a are allowed. The guide owns the
detailed steps; the SKILL.md Process is just an orientation skeleton._shared/ files (common.md for Iron Law + Report Template, plus
decay-risks.md / test-decay-risks.md as applicable) — _shared/ is NOT
auto-loaded.evals/evals.json: append with the next sequential
id at least one happy-path scenario (with the relevant risk code in
expected_output) and at least one false-positive scenario flagged
no_risk_codes: true. Each scenario needs id, name, prompt,
expected_output, mode, files.npm run validate (structure + step continuity + Process-section
presence) and npm run evals (eval schema). Fix until both pass.skills/ into
.agents/skills/ alongside the _shared symlink, then invoke it with
$<skill-name> in a Codex session and verify the Iron Law output.Codex discovers repository skills directly; no command wrapper is needed. Report the files created and the validate/evals results.
test cases passed, vs 3 / 21 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.