NIH RePORTER MCP

You can interact with NIH RePORTER data through a dedicated MCP server that lets you search NIH-funded projects and related publications in a conversational, user-friendly way. It supports narrowing results by year, investigator, organization details, funding amounts, and more, and it can return detailed project abstracts and publication information through compatible MCP clients.

How to use

To use this MCP server, you connect through an MCP-compatible client and run the server in your environment. You will then issue search commands to look up NIH-funded projects, related publications, or a combined query that returns both.

How to install

Prerequisites you need before installation:

• Python 3.12 or higher is required.

Step by step installation

# Prerequisites
# Ensure Python 3.12+ is installed

# 1) Clone the MCP server repository
git clone <repository-url>
cd mcp-nih-reporter

# 2) Create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows, use `.venv\Scripts\activate`

# 3) Install dependencies using the UV package manager
uv pip install -e .

Usage with an MCP client

Once installed, you can run the MCP server in your environment and connect with an MCP client. The server exposes commands to search NIH projects, search publications, and perform a combined search. Use your client to send those commands and receive structured results that include project titles, abstracts, PI information, organization details, and funding data.

Example client configuration

To connect from a compatible MCP client, you can configure the client to invoke the server’s runtime. The following example shows how to reference the MCP server using a local or fully qualified path to the uv runner and the server script.

Notes on server commands

The MCP server provides endpoints for searching projects, publications, and a combined search. You can also test the API connection to verify that the server is reachable from your MCP client.

Configuration and logs

Logging is enabled by default and writes to a log file in the project root. If you encounter issues, review the log file for INFO-level messages that indicate startup status, request processing, and any errors encountered by the server.

Troubleshooting

If your MCP client cannot connect, verify that the server process is running, the path to the uv runner is correct, and that your client configuration matches the command and arguments used to start the server. Check for common issues such as missing dependencies or incorrect Python environments and ensure network access if using an HTTP-based MCP endpoint.

Security and environment

Keep your environment isolated using a virtual environment and avoid exposing internal endpoints or credentials. Do not share your runtime paths or API keys in public configurations.

Additional sections

This MCP server is designed to provide flexible search capabilities for NIH RePORTER data and related publications. You can tailor result limits and refine searches by fiscal years, PI names, organization attributes, funding amounts, COVID-19 status, funding mechanisms, institute/center codes, and RCDC terms. If your workflow requires it, you can combine project and publication results to gain a comprehensive view of NIH-funded research activity.

Available tools

search_projects

Search NIH-funded research projects by criteria such as fiscal years, PI names, organization details, funding amounts, COVID-19 status, funding mechanism, and institute/center codes.

search_publications

Search publications associated with NIH projects to retrieve abstracts and related details.

search_combined

Perform a combined search that returns results from both projects and publications.

test_connection

Test the MCP API connection to ensure the server is reachable from clients.