api2mdx

api2mdx is a Python tool that generates mdx documentation for Python APIs. Under the hood, it uses Griffe.

api2mdx is designed to work in tandem with custom components like <ParametersTable>, <ApiSignature>, etc. Currently they live in @mirascope/website or @mirascope/docs-viewer, but later they will be extracted to @mirascope/ui.

Usage

Generate API documentation from Python source code:

python -m api2mdx.main --source-path ./src --package mypackage --output ./docs/api

Options

• --source-path: Path to the source code directory
• --package: Python package name to document
• --output: Path where generated documentation should be written
• --docs-path: Path within the package where docs are located (default: docs/api)
• --pattern: Optional pattern to regenerate only matching files
• --output-directives: Optional path to output intermediate directive files for debugging

Example

python -m api2mdx.main \
  --source-path ./snapshots \
  --package mirascope_v2_llm \
  --output ./snapshots/mdx \
  --output-directives ./snapshots/directives

Snapshot Regeneration

The project includes test snapshots for validation. To regenerate all snapshots:

uv run regenerate-snapshots

This will:

1. Generate MDX documentation files in snapshots/mdx/
2. Generate intermediate directive files in snapshots/directives/ (useful for debugging)
3. Process documentation links and generate metadata

Snapshot Structure

snapshots/
├── mirascope_v2_llm/        # Source Python code
├── mdx/                     # Generated MDX documentation  
└── directives/              # Intermediate directive files (for debugging)

How It Works

1. API Discovery: Scans Python modules and respects __all__ exports as the source of truth
2. Path Generation: Uses export structure for file organization (e.g., responses/__init__.py exports Response → responses/Response.mdx)
3. Directive Processing: Generates Griffe directives and processes them into MDX
4. Metadata Generation: Creates TypeScript metadata files for navigation