NIH RePORTER MCP
# NIH RePORTER MCP
A Model Context Protocol [(MCP)](https://modelcontextprotocol.io/introduction) server for chatting with [NIH RePORTER](https://reporter.nih.gov/). Search for NIH-funded research projects and publications in a conversational manner.
Accompanying blog post [here](https://open.substack.com/pub/johndamask/p/building-an-mcp-server-over-nihs?r=2ee1b&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true).

## Features
- Search NIH-funded research projects with various criteria:
- Fiscal years
- Principal Investigator names
- Organization details (name, state, city, type, department)
- Funding amounts
- COVID-19 response status
- Funding mechanism
- Institute/Center codes
- RCDC terms
- Date ranges
- Search publications associated with NIH projects
- Combined search functionality for both projects and publications
- Detailed project and publication information including abstracts
- Configurable result limits
## Prerequisites
- Python 3.12 or higher
- UV package manager (recommended for faster dependency installation)
## Installation
1. Clone the repository:
```bash
git clone <repository-url>
cd mcp-nih-reporter
```
2. Create and activate a virtual environment:
```bash
python -m venv .venv
source .venv/bin/activate # On Windows, use `.venv\Scripts\activate`
```
3. Install dependencies using UV:
```bash
uv pip install -e .
```
## Usage
This MCP server provides access to the NIH RePORTER API through several tools:
- `search_projects`: Search for NIH-funded research projects
- `search_publications`: Search for publications associated with NIH projects
- `search_combined`: Combined search for both projects and publications
- `test_connection`: Test the API connection
You can use this MCP with any MCP-compatible client, such as:
- Claude Desktop
- Cursor
- Other MCP-enabled tools
### Example claude_desktop_config.json
```
{
"mcpServers": {
"nih-reporter": {
"command": "<fully qualified path to>/uv",
"args": [
"run",
"--with",
"mcp[cli]",
"mcp",
"run",
"<fully qualified path to>/mcp-nih-reporter/mcp-nih-reporter.py"
]
}
}
}
```
The search results will be returned in a structured format containing project details including:
- Project title and abstract
- Principal Investigator information
- Organization details
- Funding information
- Project dates and status
## Debugging
A log file will be created in the root folder when the MCP attempts to run in a client (e.g. Claude Desktop). Check there if you're having trouble.
## Development
The project uses:
- `httpx` for async HTTP requests
- `mcp` for the Mission Control Protocol implementation
- `python-dotenv` for environment variable management
- `uv` for dependency management
## Logging
Logs are written to `mcp-nih-reporter.log` in the project root directory. The logging level is set to INFO by default.
## Contributing
Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.
1. Fork the repository
2. Create your feature branch (`git checkout -b feature/amazing-feature`)
3. Commit your changes (`git commit -m 'Add some amazing feature'`)
4. Push to the branch (`git push origin feature/amazing-feature`)
5. Open a Pull Request
Please make sure to update tests as appropriate and follow the existing code style.
## License
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
TDQS
Scored across 4 tools
The three search tools have overlapping purposes that could cause confusion. search_combined appears to combine projects and publications, while search_projects and search_publications handle them separately. An agent might struggle to choose between search_combined and the other two for certain queries, though the descriptions provide some guidance about their different focuses.
All tool names follow a consistent verb_noun pattern with snake_case. search_combined, search_projects, search_publications, and test_connection all use clear, predictable naming conventions that make the tool set easy to understand at a glance.
Four tools is reasonable for an NIH RePORTER interface, though it feels slightly thin. The three search operations and one connection test cover the core functionality, but additional tools for retrieving specific projects or publications by ID might have made the surface more complete. The count isn't problematic, just minimal.
The tool set provides search capabilities but lacks obvious retrieval operations. There's no way to get a specific project or publication by identifier (like project number or PMID), which creates a gap in the CRUD lifecycle. Agents can work around this by searching with specific criteria, but direct lookup tools would make the surface more complete for the NIH RePORTER domain.