rak Zotero RAG

What it does

rak is a specialized RAG (Retrieval-Augmented Generation) tool for Zotero. It indexes PDF full-texts and metadata locally using embeddings, allowing AI assistants to perform semantic searches, find similar papers, and answer complex research questions based on the user's personal academic library.

Tools

• search_papers: Performs semantic or hybrid (Vector + BM25) search and returns matching snippets.
• search_papers_bm25: Pure keyword-based search over indexed full-texts.
• similar_papers: Finds papers similar to a specific Zotero key or title.
• ask_papers: LLM-powered Q&A based on retrieved paper contexts.
• export_papers: Exports search results to CSV or BibTeX formats.
• index_status: Reports the current state of the local vector index.

Installation

Install the MCP package via pip:

pip install zotero-rag-cli[mcp]

Then add to your MCP configuration:

{
  "mcpServers": {
    "rak": {
      "command": "rak-mcp"
    }
  }
}

Supported hosts

• claude
• cursor
• zed