Skills
Built-in LLM prompt templates
Skills are LLM prompts, not code. They’re expert instructions that tell AI agents exactly how to perform a task — what to check, what to produce, and what success looks like.
Built-in Skills
| Skill | Purpose | Usage |
|---|---|---|
docsReviewAgent | Reviews doc quality page-by-page — catches stale content, missing sections, unclear explanations, broken links | Use the docsReviewAgent skill to review docs/overview.md |
promptSlideoutGenerator | Generates AI-consumable prompt configurations for documentation pages | Use promptSlideoutGenerator to create prompt config for the API page |
docsDesignCritic | Critiques page structure and visual design — heading hierarchy, component usage, information density | Use docsDesignCritic to critique docs/quickstart.md |
installMdGenerator | Creates install.md files following the installmd.org spec | Use installMdGenerator to create install.md from dewey.config.ts |
Creating Custom Skills
Skills live as markdown files in your project:
.claude/skills/
my-skill.md
Each skill follows a consistent structure:
# Skill Name
Brief description of what this skill does.
## When to Use
- Situation 1
- Situation 2
## Instructions
Step-by-step guide for the AI agent:
1. First, check X
2. Then, do Y
3. Finally, verify Z
## Example
Show an example input and expected output.
Best Practices
| Do | Don’t |
|---|---|
| Be specific and actionable | Use vague instructions |
| Include examples | Assume context |
| Define success criteria | Leave outcomes ambiguous |
| Reference file paths | Use relative descriptions |