AI Documenter

AI Documenter is an AI-powered technical documentation assistant designed to analyze codebases, files, and directories, generating accurate, well-structured documentation, README files, API docs, and technical guides. It supports both OpenAI models (cloud) and LMStudio (local) for deep code and documentation analysis, making it an ideal tool for developers and technical writers who value flexibility and privacy.

Features

• Interactive CLI: Engage in natural conversations with the documentation assistant via a terminal-based interface.
• Multiple AI Providers: Choose between OpenAI (cloud) or LMStudio (local) for AI processing.
• Automated Documentation: Generate documentation for source files, project directories, configuration files, and more.
• AI-Powered Analysis: Integrates OpenAI's latest models or local LMStudio models for advanced understanding and generation.
• Privacy-First Option: Use LMStudio for completely local processing without sending code to external APIs.
• Customizable Tools: File system read/write, directory and file info, project structure analysis, and more.
• Project-Specific Configuration: Configurable per-project with .documenter.json config files.
• Global Installation: Install once, use anywhere in your projects.
• Built-in Help & Guidance: Easily accessible help and usage examples within the CLI.

Installation

Global Installation (Recommended)

Install AI Documenter globally to use it in any project:

npm install -g ai-documenter

Local Installation

Install in a specific project:

npm install ai-documenter
npx ai-documenter

From Source

1. Clone the Repository:

   git clone <repository-url>
   cd documenter

2. Install Dependencies:

   npm install

3. Build and Install:

   npm run build
   npm link  # For global installation

Quick Start

1. Initialize AI Documenter in your project:

   cd your-project
   ai-documenter init

2. Start the documentation assistant:

   ai-documenter

3. Start documenting! Try commands like:

   • "Document the file src/app.js"
   • "Create a README for this project"
   • "Explain the src/ folder structure"

Configuration

AI Documenter supports both OpenAI (cloud) and LMStudio (local) providers for maximum flexibility.

Environment Variables

Common Settings

• LLM_PROVIDER (Optional): Choose provider: openai or lmstudio (default: openai)
• MAX_CONVERSATION_HISTORY (Optional): Number of conversation turns to remember (default: 10)
• DEFAULT_OUTPUT_DIR (Optional): Default directory for generated documentation (default: ./docs)
• LLM_TIMEOUT (Optional): Request timeout in milliseconds (default: 3600000)

OpenAI Provider

• OPENAI_API_KEY (Required for OpenAI): Your OpenAI API key
• OPENAI_MODEL (Optional): Model to use (default: gpt-4o-mini)

LMStudio Provider

• LMSTUDIO_ENDPOINT (Optional): LMStudio API endpoint (default: http://localhost:1234/v1)
• LMSTUDIO_MODEL (Optional): Model name in LMStudio (default: local-model)

Configuration File

Create a .documenter.json file in your project root (this file is automatically added to .gitignore):

OpenAI Configuration

{
  "provider": "openai",
  "openai_api_key": "your-api-key-here",
  "openai_model": "gpt-4o-mini",
  "max_conversation_history": 10,
  "default_output_dir": "./docs"
}

LMStudio Configuration

{
  "provider": "lmstudio",
  "lmstudio_endpoint": "http://localhost:1234/v1",
  "lmstudio_model": "your-local-model-name",
  "max_conversation_history": 10,
  "default_output_dir": "./docs"
}

Configuration Priority

1. .documenter.json in current directory
2. documenter.config.json in current directory
3. Environment variables
4. .env file in current directory
5. Global .env file in ~/.documenter/.env

Note: Configuration files (.documenter.json, documenter.config.json) are automatically excluded from git to prevent committing sensitive API keys. For team projects, document the required configuration in your project's README instead of committing the actual config files.

LMStudio Setup

To use LMStudio for local AI models:

1. Download and Install LMStudio from https://lmstudio.ai

2. Download a Model (e.g., Code Llama, Mistral, or other coding-focused models)

3. Start the Local Server in LMStudio (typically runs on port 1234)

4. Configure AI Documenter:

   ai-documenter init
   # Select option 2 for LMStudio when prompted

5. Set Environment Variables (optional):

   export LLM_PROVIDER=lmstudio
   export LMSTUDIO_ENDPOINT=http://localhost:1234/v1
   export LMSTUDIO_MODEL=your-model-name

Benefits of LMStudio:

• ✅ Complete privacy - your code never leaves your machine
• ✅ No API costs - run unlimited documentation generation
• ✅ Works offline - no internet connection required
• ✅ Custom models - use specialized coding models

Usage

Interactive Mode

Run ai-documenter to start the interactive documentation assistant:

ai-documenter

Initialize a Project

Set up AI Documenter configuration for your project:

ai-documenter init

Example Commands

Once in interactive mode, try these commands:

• Document the file src/app.js
• Explain the src/ folder structure
• Write API docs for this project
• Create a README for this project
• Explain how this code works
• Generate documentation for the entire codebase

CLI Commands

• exit or quit - Exit the application
• clear - Reset conversation and clear the screen
• help - Display help message
• init - Initialize documenter in current directory

Project Structure

ai-documenter/
├── src/
│   ├── cli.ts        # Main CLI application and entry point
│   ├── agent.ts      # Custom agent implementation for multiple providers
│   ├── providers.ts  # Provider abstraction (OpenAI, LMStudio)
│   ├── prompt.ts     # Agent instructions and prompt template
│   ├── tools.ts      # File and directory tools used by the agent
├── bin/              # Built executable (generated)
├── package.json      # Project configuration and dependencies
└── README.md         # This file

Configuration Examples

Basic Setup

1. Global API Key (works everywhere):

   echo "OPENAI_API_KEY=your-key-here" >> ~/.zshrc  # or ~/.bashrc
   source ~/.zshrc

2. Project-Specific Setup:

   cd your-project
   ai-documenter init
   # Follow the prompts to configure

Advanced Configuration

Create .documenter.json for project-specific settings:

{
  "openai_model": "gpt-4o",
  "default_output_dir": "./documentation",
  "max_conversation_history": 20
}

Development

Running from Source

npm run dev

Building

npm run build
./bin/ai-documenter

Publishing

npm run build
npm publish

Customization

• Edit src/prompt.ts to change instructions for the technical writer agent
• Enhance src/tools.ts to extend or modify file system tools
• Modify src/cli.ts to add new commands or change behavior

Use Cases

• API Documentation: Generate comprehensive API docs from source code
• README Files: Create project overviews and setup instructions
• Code Explanations: Understand and document complex codebases
• Technical Guides: Create step-by-step implementation guides
• Architecture Documentation: Document system design and structure
• Configuration Documentation: Explain config files and settings

Tips

• Use specific file paths for targeted documentation
• Ask for explanations of code patterns and architecture
• Request documentation in specific formats (Markdown, HTML, etc.)
• Combine multiple requests in conversation for comprehensive docs
• Use the clear command to start fresh documentation sessions

Troubleshooting

API Key Issues

If you get an API key error:

1. Run ai-documenter init to set up configuration
2. Check environment variables: echo $OPENAI_API_KEY
3. Verify .env file exists in your project or home directory
4. Check .documenter.json configuration file

Permission Issues

If you get permission errors:

# Fix npm permissions
npm config set prefix ~/.npm-global
export PATH=~/.npm-global/bin:$PATH

Build Issues

If the build fails:

# Clean and rebuild
rm -rf bin/
npm run build

Contributing

Pull requests, bug reports, and feature suggestions are welcome! Please open an issue for discussion before making major changes.

License

MIT License