SKILL.md Template

---
name: [skill-name-in-kebab-case]
description: >
  [What the skill does in third person, then WHEN to use it, including the
  exact words users type. E.g.: "Generates release notes from merged PRs.
  Use when the user asks for release notes, a changelog, or 'what shipped',
  or before tagging a release."]
---

# [Skill Name]

[2-3 sentences: what this skill accomplishes and the situation it's for.]

## When to use

Use this skill when: [the concrete triggers: user phrases, file types, task shapes].
Do NOT use it for: [the adjacent task people confuse it with: point to the right approach instead].

## Workflow

1. [First step, start with any information-gathering: which files to read, what to ask the user if missing.]
2. [Core steps, one action each. Reference helper files by relative path, e.g. "run `scripts/validate.py` on the output".]
3. [Verification step, how the agent checks its own work: run the validator, compare against the checklist, re-read the output against the rules below.]
4. [Delivery step: exact format of what to return.]

## Rules

- [Hard constraint, stated positively: "Always X because Y."]
- [The mistake this skill exists to prevent, as an explicit rule.]
- If [common edge case], then [what to do instead].

## Example

Input: [a realistic user request]
Output:
```
[a short, exactly-formatted example of the desired result]
```

## References

- `references/[topic].md`: read when [specific condition, e.g. "the user asks for sources"].
- `scripts/[helper].[ext]`: [what it does; when to run it].