Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →sweetcornna/mathodology-skill-authoring
Use when adding, editing or reviewing skills, role prompts or skill metadata.
Passed 17 of 22 test cases with the skill, and 5 of 22 without it.
and 15 more in the full report
12 fixed5 passed either way5 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 12 test cases the skill fixed. Same model, same prompt, with and without the skill.
I am drafting a new skill in a directory named 'kubernetes-canary-deployments'. Produce the frontmatter for SKILL.md.
| Checked | Without skill | With skill |
|---|---|---|
| Set the frontmatter name attribute to kubernetes-canary-deployments. | Passed | Passed |
| Start the frontmatter description attribute with "Use when". | Missed | Passed |
The instructions your model reads when the skill loads.
Write reusable guidance around decisions, evidence and useful outputs. Explain when a method helps, how to recognize failure and when a simpler approach is better. Avoid mandatory phases, handoff schemas, arbitrary scores and prose that merely repeats another skill. Link the owning skill instead.
Every skill has SKILL.md frontmatter with a name equal to its directory and
a trigger-focused description starting with Use when. Its agents/openai.yaml
contains display_name, short_description and a default_prompt mentioning the
matching $skill-name. These metadata files are discovery interfaces, not a
workflow engine. Use concise role definitions; inherit the host's model choice.
Put detailed figure recipes, source provenance and examples in references or examples directories within the owning skill. Keep the entry point short and load only the references relevant to the task. Preserve third-party licensing; never execute downloaded source snapshots as an installation step.
Small scripts are appropriate when they perform concrete work such as plotting, backup or rendering. Keep them optional, declare dependencies, and verify the behavior they actually promise. Do not create a framework to enforce prompts.
Update all affected entry points, role skill lists, metadata and English/Chinese documentation. Keep each language natural; heading or code-block counts need not match. Check that commands, capabilities and links agree. Use maintenance for mechanical checks.
test cases passed, vs 5 / 22 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.