Your Skill Name

One sentence saying what this skill does and what it produces. Then the triggers — Use when the user says "phrase one", "phrase two", "phrase three", or asks to <do the thing in their own words>. This description is the ONLY text Claude reads when deciding whether to use the skill, so the trigger phrases live here, not in the body.

wavebound-ai 9b1a786 1.3 KB Updated

File contents

Skill Title

One short paragraph: who invokes this, what situation they're in, and what outcome they want. Write it so the model understands the intent, not just the task.

Rules

A few strong constraints beat many weak ones. Bold the first sentence of each rule, then explain it in one or two more.

The most important constraint goes first. Usually: what the skill must NOT do (scope it), or what must always lead the output (order it).

A second rule. For example: tone, honesty requirements, what to do when information is missing.

Output

Exactly what the response should look like — format, length cap, required sections, and what to omit. The more concrete this section is, the more consistent the results. Example shape:

Section one — what goes here, how long.

Section two — what goes here. Omit entirely when it doesn't apply (say so).

End with the single behavior that keeps the skill sharp, e.g. "Commit to one recommendation; never offer a menu."

wavebound-ai/wavebound-skills/tree/main/templates/skill-template commit 9b1a786842

Frequently asked questions

npx skillmds@latest add wavebound-ai/your-skill-name