NRTSearch

NRTSearch is an MCP server that connects Lucene-based search indexes powered by NRTSearch to AI assistants such as GitHub Copilot and Claude Desktop. Developed with modern Python packaging and Pydantic, the server exposes full-text search capabilities over Model Context Protocol, allowing language models to query backend indexes directly during conversations. Search engineers, software developers, and research teams use it to retrieve indexed documents, analyze search rankings, and test query syntax using natural language. The server supports comprehensive Lucene query parsing, including Boolean operators, exact phrases, range queries, wildcards, and fuzzy matching. It provides configurable hit limits, response field selections, hit highlighting, retry logic, and structured logging. By translating conversational context into structured index queries, NRTSearch enables AI models to surface internal knowledge bases, e-commerce inventories, or domain-specific search records without requiring manual query construction in external terminal sessions.

Category: Databases & Data Stores

Tags: indexing, information-retrieval, lucene, search

Visit NRTSearch

How to install and configure NRTSearch

1. Clone the repository and navigate to the directory: bash git clone https://github.com/tvergilio/nrtsearch-mcp-server.git cd nrtsearch-mcp-server 2. Run the quickstart script to install dependencies and configure the environment: bash ./quickstart.sh 3. Alternatively, launch the server manually using the Python module or CLI: bash python -m nrtsearch_mcp.server # or nrtsearch-mcp 4. Configure your connection settings via environment variables like LOG_LEVEL or by pointing NRTSEARCH_MCP_CONFIG to a JSON configuration file containing your nrtsearch_connection host and port. 5. In your client (such as VS Code for GitHub Copilot or Claude Desktop), register the server by referencing the execution command or path according to the repository documentation at https://github.com/tvergilio/nrtsearch-mcp-server.

What you can do with NRTSearch

• Running complex Lucene queries through AI prompts using Boolean syntax, range filters, and fuzzy matching. - Inspecting matching text snippets with keyword highlighting to verify document relevance across search indices. - Integrating enterprise search indices with GitHub Copilot to surface relevant codebase documentation or knowledge bases. - Testing and debugging Lucene query behaviors by retrieving customized field sets and score breakdowns directly within chat sessions.

Key facts

• Open Source
• https://github.com/tvergilio/nrtsearch-mcp-server
• Databases & Data Stores, Web Search & Research
• indexing, information-retrieval, lucene, search

Part of MCP Servers

How do I install NRTSearch?

Clone the repository using git, switch into the project directory, and execute the provided quickstart script to automatically install all dependencies. You can also run the server directly using Python module commands or via the nrtsearch-mcp command-line interface.

What can NRTSearch do?

NRTSearch exposes the nrtsearch/search tool to Model Context Protocol clients. It accepts index names, Lucene query strings, top hit limits, field retrieval lists, and highlighting preferences, returning ranked search hits directly to your AI assistant.

Which MCP clients work with NRTSearch?

NRTSearch supports standard MCP clients, including GitHub Copilot in VS Code and Claude Desktop. Any client compatible with the Model Context Protocol can connect to the server once configured in the client settings.

Is NRTSearch open source?

Yes, NRTSearch is open-source software distributed under the Apache License 2.0. The source code and licensing details are available in the project repository on GitHub.