AutoDocGen

AI-powered Markdown documentation generator for Python projects.

AutoDocGen scans your Python source code, extracts docstrings and structure using Python's built-in AST parser, and uses AI to generate comprehensive, readable Markdown documentation — one file per module plus a navigable index.

Features

• Multi-Provider Support — Works with OpenRouter, Groq, Google Gemini, and OpenAI.
• AST-based parsing — No fragile regex. Uses Python's ast module to accurately extract modules, classes, methods, and functions.
• Jinja2 Templates — Clean, customizable Markdown output using industry-standard templates.
• Async function support — Correctly identifies and labels async def functions.
• Cross-module linking — Automatically traces imports to create related module links.
• Navigable index — Produces a comprehensive index.md.
• Flexible configuration — Supports .env, autodocgen.yaml, pyproject.toml, and CLI flags.

Installation

git clone https://github.com/shifulegend/autodocgen
cd autodocgen
pip install .

Configuration

Set your API keys in a .env file (which is gitignored):

OPENAI_API_KEY=sk-...
GROQ_API_KEY=gsk_...
OPENROUTER_API_KEY=sk-or-...
GOOGLE_API_KEY=AIza...

autodocgen.yaml example

llm:
  provider: groq
  model: llama-3.1-8b-instant

output:
  dir: docs

source:
  - autodocgen

Usage

# Generate docs using settings from autodocgen.yaml
autodocgen . --verbose

# Override provider/model via CLI
autodocgen . --output docs_custom -v

Roadmap

• Multi-provider LLM support (OpenRouter, Groq, Gemini)
• Jinja2 template-based rendering
• Async AI calls for faster generation
• Incremental generation (skip unchanged files)
• HTML and reStructuredText output formats

License

MIT — Copyright (c) 2025-2026 AutoDocGen Team