MCP RAG Server

MCP RAG Server is an MCP server that connects generative AI applications and Model Context Protocol clients to retrieval-augmented generation pipelines and external scientific literature repositories. Built using FastMCP, LangGraph adapters, and AWS services, it allows AI agents to query external databases such as arXiv, PubMed, ChEMBL, and ClinicalTrials.gov, alongside Amazon Bedrock Knowledge Bases. Developers, AI researchers, and life science practitioners use this server to equip language models with up-to-date document retrieval, multi-source reference capabilities, and domain-specific knowledge ingestion. By bridging AI agents with specialized databases and AWS serverless infrastructure, it enables conversational systems to ground their answers in real research papers, medical trials, and proprietary enterprise documentation via standard JSON-RPC communication.

Category: AI Memory & Context

Tags: arxiv, aws-lambda, pubmed, rag, retrieval

Visit MCP RAG Server

How to install and configure MCP RAG Server

1. Install the required Python dependencies by running pip install mcp langchain-mcp-adapters in your environment. 2. Clone the repository and configure your AWS credentials to enable access to Amazon Bedrock and Knowledge Bases. 3. Configure your MCP client (such as Claude Desktop, Cursor, or a custom LangGraph agent) by registering the server script under mcpServers using standard stdio parameters. 4. Point the command to your Python executable and pass the target server script (such as application/mcp_server_arxiv.py or application/mcp_server_pubmed.py) in the arguments list. 5. Restart your MCP host client to establish the connection and discover the retrieval tools.

What you can do with MCP RAG Server

• Search academic papers on arXiv to retrieve research abstracts, authors, and findings directly from an agent interface. - Query PubMed and ClinicalTrials.gov to fetch clinical studies, medical evidence, and drug discovery documentation. - Retrieve enterprise internal context from Amazon Bedrock Knowledge Bases using managed vector search and semantic chunking. - Integrate literature retrieval tools into LangGraph or Strands multi-agent workflows for automated scientific research synthesis.

Key facts

• Open Source
• https://github.com/kyopark2014/mcp
• AI Memory & Context, Web Search & Research
• arxiv, aws-lambda, pubmed, rag, retrieval

Part of MCP Servers

How do I install MCP RAG Server?

You install the core dependencies by running pip install mcp langchain-mcp-adapters. Afterwards, set up your AWS credentials for Amazon Bedrock and configure the specific server script inside your MCP client JSON configuration using the stdio transport.

What can MCP RAG Server do?

The server exposes search tools that retrieve relevant data from scientific and clinical sources including arXiv, PubMed, ChEMBL, and ClinicalTrials.gov. It also interfaces with Amazon Bedrock Knowledge Bases to perform vector search over external organizational documents.

Which MCP clients work with MCP RAG Server?

It functions with any MCP-compliant client that supports stdio or SSE transport. This includes Claude Desktop, Cursor, Streamlit agent interfaces, and custom agentic frameworks built with LangGraph, Strands, or the Claude Agent SDK.

Is MCP RAG Server open source?

Yes, MCP RAG Server is an open-source project hosted on GitHub under the repository kyopark2014/mcp, making its code freely available for developers to inspect, modify, and extend.