skrypt

AI-powered documentation generator with code examples.

Scans your codebase, extracts API signatures, and generates beautiful documentation with working code examples.

Features

• Multi-language scanning: TypeScript, Python, Go, Rust
• AI-generated docs: Clear descriptions, parameters, return types
• Code examples: Working examples in TypeScript and Python
• Topic organization: Group elements by concept, not file
• MDX output: Works with any doc platform
• Doc site template: Beautiful, searchable documentation site

Installation

npm install -g skrypt-ai

Quick Start

1. Initialize a documentation site

skrypt init my-docs
cd my-docs
npm install

2. Generate API documentation

skrypt generate ../my-project/src -o ./content/docs

3. Start the dev server

npm run dev

Open http://localhost:3000 to see your docs.

CLI Commands

skrypt init [directory]

Initialize a new documentation site.

skrypt init my-docs --name "My API Docs"

Options:

• --name <name> - Project name (default: "my-docs")

skrypt generate <source>

Generate documentation from source code.

skrypt generate ./src -o ./docs

# Multi-language examples
skrypt generate ./src -o ./docs --multi-lang

# Organize by topic
skrypt generate ./src -o ./docs --by-topic

Options:

• -o, --output <dir> - Output directory
• -c, --config <file> - Config file path
• --provider <name> - LLM provider (deepseek, openai, anthropic, google, ollama, openrouter)
• --model <name> - LLM model name
• --multi-lang - Generate TypeScript + Python examples
• --by-topic - Organize by topic instead of file
• --dry-run - Scan only, don't generate
• --public-only - Only document exported/public APIs
• --exclude <patterns...> - Exclude patterns (files or name:pattern)
• --llms-txt - Generate llms.txt for Answer Engine Optimization
• --project-name <name> - Project name for llms.txt header
• --openapi <file> - Include OpenAPI spec for API Playground

Privacy Controls

Prevent internal code from being documented:

# Only document exported APIs
skrypt generate ./src --public-only

# Exclude specific patterns
skrypt generate ./src --exclude "**/internal/**" "name:_private*"

Create .skryptignore in your source directory:

# Exclude test files
**/*.test.ts
**/*.spec.ts

# Exclude internal modules
**/internal/**

# Exclude by function name (regex supported)
name:_privateHelper
name:__internal.*

Answer Engine Optimization (AEO)

Generate llms.txt for AI search engines:

skrypt generate ./src --llms-txt --project-name "My API"

This creates:

• llms.txt - Condensed API summary for LLMs
• llms-full.md - Full API reference in LLM-friendly format

Configuration

Create .skrypt.yaml in your project:

version: 1

source:
  path: ./src
  include: ["**/*.ts", "**/*.py"]
  exclude: ["**/node_modules/**"]

output:
  path: ./docs
  format: mdx

llm:
  provider: deepseek
  model: deepseek-v3

Doc Site Features

The generated documentation site includes:

• Search: Full-text search across all docs
• Syntax highlighting: Beautiful code blocks with Shiki
• MDX components: Cards, Tabs, CodeGroup, Callouts, Accordions, Steps
• Dark mode: Automatic dark/light theme switching
• Responsive: Works on all device sizes

MDX Components

Use these components in your MDX files:

<CardGroup cols={2}>
  <Card title="Getting Started" icon="Zap" href="/docs/quickstart">
    Get up and running in 5 minutes
  </Card>
</CardGroup>

<Info title="Note">
  Important information here.
</Info>

<CodeGroup>
\`\`\`typescript
const client = new Client()
\`\`\`

\`\`\`python
client = Client()
\`\`\`
</CodeGroup>

<Steps>
  <Step title="Install">Install the package</Step>
  <Step title="Configure">Set up your config</Step>
</Steps>

Environment Variables

Set API keys for your LLM provider:

export DEEPSEEK_API_KEY=your-key
export OPENAI_API_KEY=your-key
export ANTHROPIC_API_KEY=your-key