AI-powered TypeScript docstring generator. Add JSDoc comments to your TypeScript files using local AI or manual input.

• Scans all .ts and .tsx files recursively from your current directory
• Interactive selection with arrow key navigation and fuzzy search
• AI-powered generation using local Ollama, or manual input as fallback
• Multiple templates for different documentation styles
• Safe operations with automatic backups before any changes

npm install -g docki

docki

• Node.js 16+
• TypeScript files (.ts, .tsx)
• Ollama (optional, for AI generation)

Docki now includes 7 specialized templates designed for different types of TypeScript files, with intelligent auto-detection based on file content and location.

The ai-function template is specifically designed for AI/ML code:

/**
 * @description Generates embeddings for text using OpenAI API
 * 
 * @param {string} text - The input text to embed
 * @param {string} model - OpenAI model to use (default: text-embedding-ada-002)
 * @returns {Promise<number[]>} Array of embedding vectors
 * 
 * @example
 * const embeddings = await generateEmbeddings("Hello world", "text-embedding-ada-002")
 * console.log(embeddings.length) // 1536
 * 
 * @author AI Team
 */

Docki automatically suggests the best template based on:

• File location: /api/ → api-endpoint, /ai/ → ai-function, /components/ → react-component
• File naming: useAuth.ts → react-hook, aiModel.ts → ai-function
• Content analysis: Detects React components, API handlers, AI/ML keywords

All templates support these tokens (unused tokens are automatically removed):

You can create custom templates by adding .txt files to the templates/ directory:

my-custom.txt:

/**
 * @description {{description}}
 * 
 * @version 1.0.0
 * @since {{date}}
 * @param {{params}}
 * @returns {{returns}}
 * 
 * {{author}}
 */

Configuration with template mapping:

{
  "templateMap": {
    "api-endpoint": "my-custom",
    "react-component": "comprehensive"
  }
}

This maps specific file types to your preferred templates.

Docki supports multiple AI providers for automatic documentation generation:

Use Google's Gemini AI for cloud-based documentation generation:

# Set your API key (get one free at https://makersuite.google.com/app/apikey)
export GEMINI_API_KEY="your-api-key-here"

# Run Docki - it will automatically detect and use Gemini
docki

Use Ollama for completely offline, private documentation generation:

# Install Ollama from https://ollama.ai
# Pull a model
ollama pull codellama:7b

# Run Docki
docki

Docki automatically detects available AI providers in this order:

1. Gemini - If GEMINI_API_KEY is set
2. Ollama - If installed locally
3. Manual - Falls back to manual input if no AI is available

Create a .docstring-cli.json file in your project root:

{
  "extensions": [".ts", ".tsx", ".js", ".jsx"],
  "excludeDirectories": ["node_modules", "dist", ".git"],
  "defaultTemplate": "jsdoc",
  "defaultAuthor": "Your Name",
  "aiProvider": "gemini",
  "geminiModel": "gemini-1.5-flash",
  
  "ollamaModel": "codellama:7b",
  "maxDescriptionLength": 80
}

• extensions: File extensions to scan for (default: TypeScript files)
• excludeDirectories: Directories to ignore during scanning
• defaultTemplate: Template to use by default
• defaultAuthor: Your name to include in docstrings
• aiProvider: Preferred AI provider: "gemini", "ollama", or "none" (auto-detect if not set)
• geminiModel: Gemini model to use (default: "gemini-1.5-flash")
• geminiApiKey: Your Gemini API key. Strongly recommended to use the GEMINI_API_KEY environment variable instead of storing it directly in this file.
• ollamaModel: Ollama model to use (default: "codellama:7b")
• maxDescriptionLength: Maximum line length for AI-generated descriptions (default: 80)

MIT © Remco Stoeten