Chewed

Code documentation toolkit for analyzing and generating LLM-friendly, but human-readable and lovable code documentation.

Features

• Code structure analysis using Python's AST
• Configurable documentation generation (MyST/Markdown)
• Automatic type annotation resolution
• Cross-component relationship mapping
• Example extraction from docstrings and tests
• Flexible configuration via pyproject.toml

Roadmap

• LLM-assisted documentation analysis & generation
• LLM-assisted documentation testing
• LLM-assisted documentation validation
• LLM-assisted documentation deployment

Installation

# Install with pip substitute
python3 -m pip install uv
uv pip install git+https://github.com/puroman/chewed.git

# Editable install for contributors
git clone https://github.com/puroman/chewed.git
cd chewed && uv pip install -e .

Get Started

# Generate docs for current project (default output dir)
chew . --output docs/

# Analyze specific package
chew ./my_module --output docs/ --verbose
chew requests --output docs/requests --local

Research Notes

from chewed import analyze_package, generate_docs

# Experimental analysis pipeline
results = analyze_package("mypackage")
generate_docs(results, output_format="myst")

Note: chewed is a research prototype - interfaces may evolve as we explore new documentation paradigms, LLM-assisted workflows, and agentic automation.

Configuration

Add to pyproject.toml:

[tool.chewed]
output_format = "myst"
exclude_patterns = ["tests/*"]
known_types = { "DataFrame" = "pandas.DataFrame" }

Setting	Description
output_format	Documentation format (myst/markdown)
exclude_patterns	File patterns to ignore
known_types	Type annotation simplifications

Project Structure

chewed/
├── src/
│   └── chewed/       # Core research implementation
│       ├── analysis/  # AST processing components
│       └── formats/   # Output format handlers
├── tests/             # Experimental validation
└── pyproject.toml

Contributing

We welcome collaborations expecialy from research teams! We would like to work on the following topics:

• Benchmarking methodologies
• Documentation patterns research
• Experimental design principles
• LLM-assisted and LLM-led documentation workflows
• Natural language documentation generation
• On-demand documentation generation
• Usage examples generation

Please see our contribution guidelines for more details.

---

MIT Licensed | Part of ongoing research into API documentation systems and agentic workflow automation