mcp-otzaria

This MCP server exposes a Jewish text search capability. It provides a single MCP tool called full_text_search that lets you perform full-text searches across a Jewish library with a powerful, query-rich syntax. You can filter by fields such as text, reference, and topics, combine terms with Boolean operators, require or exclude terms, and search exact phrases or with wildcards. The results return reference information, relevant topics, highlighted excerpts, and a relevance score to help you locate the most pertinent passages. To use it, connect via an MCP client (for example Claude or other clients that speak Model Context Protocol) and invoke the full_text_search tool with your query. The server is run through uv (Python) and reads its data from a designated directory where the library index is stored.

Prerequisites:

• Python 3.10 or higher
• Internet access to install dependencies

1. Clone the repository and navigate into it:

git clone https://github.com/sivan22/mcp-otzaria-server.git
cd mcp-otzaria-server

2. Prepare the index:

• Download and extract the index file from the provided link (the index contains the Jewish texts to be searched).

3. Install Python dependencies:

pip install .

4. Run the server locally (examples):

uv --directory /path/to/your/index/directory run jewish_library

5. Optional: configure MCP clients to point to the server using the provided mcpServers configuration snippet.

Tips and notes:

• Ensure Python 3.10+ is installed and the index is downloaded before starting the server.
• The server uses MCP SDK and Tantivy; make sure dependencies are satisfied by running the install step above.
• If you encounter encoding issues with output, the recommended environment variable is PYTHONIOENCODING=utf-8.
• The mcp_config example uses the directory where the index resides; replace path/to/directory with your actual path.
• The single tool exposed is full_text_search; queries support fielded searches (text:, reference:, topics:), boolean operators, required/excluded terms (+term, -term), phrase searches, and wildcards.