.agents/skills/skill-creator/references/templates.md
Use these templates as starting points. Keep them concise and agent-agnostic.
SKILL.md template---
name: <skill-name>
description: <what this skill does and when to use it>
license: MIT
---
# <Skill Title>
## Goal
<one paragraph describing expected outcomes>
## Procedure
1. <step 1>
2. <step 2>
3. <step 3>
## Inputs
- <input 1>
- <input 2>
## Outputs
- <output contract>
## Edge cases
- <edge case 1 and handling>
- <edge case 2 and handling>
## Validation
- [ ] Frontmatter is valid YAML
- [ ] Name matches folder
- [ ] Description includes trigger contexts
- [ ] Relative file references resolve
<skill-name>/
├── SKILL.md
├── scripts/
│ └── <script>.py
├── references/
│ ├── <domain>.md
│ └── <schemas>.md
└── assets/
└── <template-or-static-file>
Use this expanded layout only if those resources are truly needed.
Write descriptions that include both capability and activation cues.
Pattern:
"<Core capability>. Use when users ask to <intent 1>, <intent 2>, or <intent 3>, including phrasing such as <example cues>."
Example:
"Generate release notes from Git history with categorized changes. Use when users ask to draft changelogs, summarize commits for releases, or prepare version notes from tags or commit ranges."
When improving an existing skill:
SKILL.md and user complaints.description for better trigger clarity.SKILL.md to references/.references/spec-checklist.md.