A Python interface to the Mu2e docdb with support for document retrieval, vector search, and LLM chat interface.
For installation and setup instructions see here: doc/install.md
The package provides a command-line tool mu2e-docdb with the following commands (just run mu2e-docdb for help):
- List documents from the last day
mu2e-docdb list - Generate embeddings for vector search:
mu2e-docdb generate --days 10 - Vector search to retrieve relevant documents:
mu2e-docdb search "mu2e timeline" - Full-text keyword search:
mu2e-docdb search "timeline" --fulltext - Use Argo embeddings:
mu2e-docdb --argo search "detector design"See doc/docdb.md for more details.
LLM based chat interface to docdb (run mu2e-chat --help for help):
- starting an interactive mode (allowing threads) by typing
mu2e-chat(CTRL-C to quit) - single question mode:
mu2e-chat "\\rag What is the latest timeline?"
This commands runs a docdb chat with a slack interface. For more details see doc/slack.md.
Model Context Protocol (MCP) server for integration with Claude Desktop and other MCP-compatible AI assistants. Provides vector search, full-text search, and document retrieval tools. See mu2e/mcp/docdb/REDME.md for setup instructions.
Web interface providing document search, chat, and document lookup through a browser. Run with mu2e-web and access at http://localhost:5000. See doc/web.md for details.
Provides tools for benchmarking database and evaluating retrieval performance. See doc/eval.md.