Skill File Format

---
# YAML Frontmatter
name: Skill Name
description: One-sentence purpose
tools: [bash, read, write]
tags: [tag1, tag2]
category: certified
author: author-name
source: org/repo-name
---

## Overview
What this skill does and when to use it.

## Steps
Detailed procedure the agent should follow.

## Rules
Constraints and non-negotiables.

## Examples
Concrete scenarios with expected inputs and outputs.

## Overview
This skill creates PDF documents from HTML content using Puppeteer.
Use it when a task requires generating downloadable PDF output.
Requires Node.js 18+ installed on the machine.

## Steps
1. Read the source content (HTML, Markdown, or structured data)
2. Set up the HTML template with proper styling
3. Launch a headless browser using Puppeteer
4. Render the HTML and generate the PDF
5. Save the PDF to the specified output path
6. Verify the PDF opens correctly

## Rules
- Always use A4 page size unless the task specifies otherwise
- Include page numbers in the footer
- Never embed external fonts — use system fonts only
- Maximum file size: 10MB
- Always set `printBackground: true` for styled documents

## Examples

### Invoice generation
Input: JSON invoice data with line items
Output: Formatted PDF invoice with company header, line items table, and total

### Report export
Input: Markdown report with charts (as images)
Output: Multi-page PDF with proper page breaks between sections

~/.claude/skills/
└── my-skill/
    └── SKILL.md

~/.exponential/registry/skills/{skill-id}/SKILL.md

---
name: Code Documentation
description: Generates comprehensive code documentation from source files
tools: [read, write, glob, grep]
tags: [documentation, code, jsdoc]
author: marc
---

## Overview

This skill generates documentation for codebases. It reads source files,
extracts function signatures, types, and comments, then produces structured
markdown documentation.

Best used for: API documentation, library reference docs, internal code guides.

## Steps

1. Use `glob` to find all source files matching the target patterns
   (e.g., `src/**/*.ts`, `lib/**/*.py`)
2. Read each file and extract:
   - Exported functions and their signatures
   - Type definitions and interfaces
   - JSDoc/docstring comments
   - Module-level documentation
3. Organize by module/directory structure
4. Generate markdown documentation with:
   - Table of contents
   - Module overview sections
   - Function reference tables (name, params, return type, description)
   - Type definitions
   - Usage examples from existing comments
5. Write the documentation to the specified output file

## Rules

- Only document exported/public APIs — skip internal/private functions
- Preserve the original directory structure in the documentation hierarchy
- If a function has no documentation, note it as "Undocumented" — don't invent docs
- Use fenced code blocks for all code examples
- Link between related types and functions using markdown anchors
- Generate a table of contents for documents with more than 5 sections

## Examples

### TypeScript library

Input: `src/` directory with TypeScript modules
Output: `docs/API.md` with:
- Module listing
- Function signatures with parameter types
- Interface definitions
- Usage examples from JSDoc @example tags

### Python package

Input: `mypackage/` directory with Python modules
Output: `docs/reference.md` with:
- Module listing
- Function signatures with type hints
- Class definitions with method tables
- Docstring content preserved as descriptions