Attribution
This project was extracted from pw-mcp on December 25, 2025.
Origin
The code in this repository was originally developed as part of the pw-mcp project (ProleWiki MCP Server), which provides semantic vector search over the ProleWiki corpus.
The AI training components were separated into this standalone repository to:
- Reduce the dependency footprint of pw-mcp
- Allow independent versioning and release cycles
- Separate the ~1GB ChromaDB releases from the ~4-16GB model releases
- Enable focused development on training infrastructure
Original Authors
The original code was developed by contributors to the pw-mcp project.
Files Extracted
The following files were copied from pw-mcp:
src/pw_mcp/ai_training/*.py→src/prolewiki_llm/training_data/(entire directory)tests/unit/training/→tests/unit/ai-docs/(training-related YAML files)
License
This project maintains the same AGPL-3.0-only license as the original pw-mcp project.