DocAuto

A tool for generating documentation from code using LLMs.

CLI Usage

DocAuto provides a command-line interface for generating documentation from your code. The CLI is implemented in cli.py and offers several commands and options.

Basic Usage

python -m docauto [paths...] [options]

Preset Configurations

DocAuto provides preset configurations for popular LLM services:

Ollama Preset

python -m docauto --ollama [paths...]

Uses the following configuration:

• Base URL: http://localhost:11434/v1
• Model: phi4
• Max Context: 16384

OpenAI Preset

python -m docauto --openai -k your_api_key [paths...]

Uses the following configuration:

• Base URL: https://api.openai.com/v1
• Model: gpt-4o-mini
• Max Context: 16384
• Requires API key

Gemini Preset

python -m docauto --gemini -k your_api_key [paths...]

Uses the following configuration:

• Base URL: https://generativelanguage.googleapis.com/v1beta/openai/
• Model: gemini-2.0-flash-exp
• Max Context: 131072
• Requires API key

Options

API Configuration

• -b, --base-url: API base URL for the LLM service
• -k, --api-key: API key for authentication
• -m, --model: AI model to use
• -mc, --max-context: Maximum context size
• -c, --constraint: Add a documentation constraint (can be used multiple times)
• -d, --dry-run: Show changes without modifying files
• -v, --verbose: Enable verbose logging
• -o, --overwrite: [Dangerous] Overwrite existing docstrings in codebase
• paths: One or more Python files or directories to process

Examples

# Process a single file
python -m docauto example.py

# Process multiple files with API key
python -m docauto file1.py file2.py -k your_api_key

# Process a directory in dry-run mode
python -m docauto ./src -d

# Process with custom model and verbose logging
python -m docauto ./src -m gpt-4 -v

# Use Ollama preset for local processing
python -m docauto ./src --ollama

# Use OpenAI preset with API key
python -m docauto ./src --openai -k your_api_key

The tool will process the specified Python files, generate documentation using the configured LLM, and update the files with the generated documentation. Use the --dry-run option to preview changes without modifying files.