How It Works
- You specify a library (e.g., React, FastAPI, Prisma)
- Exa semantic search finds documentation, best practices, and patterns for that library
- LLM processes the gathered information into a structured .mdc rule file
- Output is a ready-to-use Cursor rule file tailored to that specific library
Usage
Generate for a specific library
uv run src/generate_mdc_files.py --library fastapiGenerate for a category
uv run src/generate_mdc_files.py --tag backendBatch generate with concurrency
uv run src/generate_mdc_files.py --workers 4 --rate-limit 10Regenerate all rules
uv run src/generate_mdc_files.py --regenerate-allAdding New Libraries
Edit rules.json:
{
"name": "your-library",
"tag": "category"
}Then run the generator to create the .mdc file automatically.
Requirements
- Python 3.8+
- Exa API key (for semantic search)
- LLM API key (Gemini, OpenAI, or Anthropic)