Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →fedoroff-vlad/new-skill
Use when you are about to author a NEW reusable skill in this repo, or rewrite an existing skill's frontmatter so it triggers reliably. Fires on: "create a skill", "add a skill", "why doesn't my skill fire", "write a SKILL.md", "make this skill discoverable". Produces a correctly-shaped SKILL.md with a trigger-style description and updates the skills index.
Passed 19 of 21 test cases with the skill, and 5 of 21 without it.
and 19 more in the full report
15 fixed5 passed either way2 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 14 test cases the skill fixed. Same model, same prompt, with and without the skill.
Show the exact line to append to `SKILLS-INDEX.md` for a newly created skill `lint-helm-charts` that checks Helm charts when pushing to staging.
| Checked | Without skill | With skill |
|---|---|---|
| Formats line using `lint-helm-charts — Use when...` pattern | Missed | Passed |
The instructions your model reads when the skill loads.
You are creating a portable skill for this repo. A skill is prompt + I/O
contract, not an agent: it owns no state. Provider-neutral by rule — the body
must read the same to Claude Code, to a custom SkillRegistry loader, or to a
bare LLM that gets the body injected as context. Do not name host-specific tools.
A skill is selected because its description sits in context (level-1) and the
model matches the user's request against it. So write the description as a
"use when …" statement, not a "this does X" statement.
Checks documentation consistency.Use when you changed an env var / port / tool / dependency and are preparing a PR — verifies every coupled artifact moved.Embed concrete trigger phrases, in every language the users speak, when the
skill is user-invoked (e.g. "разбери инбокс" / "help me clarify my inbox").
triggers: []): selected by an LLM classifier over the
description. The description is load-bearing — invest in "use when".triggers: [some.kind]): selected by a deterministic
string match on an event kind, no LLM. Here the triggers: list IS the
"use when"; the description is documentation only. Do not add "use when"
phrasing to a trigger skill — it is never read for selection.Only the description is always in context. The body loads when the skill runs.
So: put the trigger surface in the description; put the full procedure in the
body; push long reference material into sibling files the body links to and the
agent opens on demand. A bloated body is a tax paid on every invocation.
description last, as "usetest cases passed, vs 5 / 21 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.