
Hugging Face MCP Server
MCP server providing access to the Hugging Face Dataset Viewer API. Query datasets, explore data, search content, and analyze statistics from the Hugging Face Hub's extensive collection of machine learning datasets.
Features
- 12 MCP Tools covering all API endpoints:
get_dataset_splits- Dataset splits informationget_dataset_info- Dataset metadata and featuresget_dataset_first_rows- Preview first 100 rowsget_dataset_rows- Paginated data accesssearch_dataset- Full-text search within datasetsget_dataset_size- Dataset size informationget_dataset_statistics- Statistical analysisfilter_dataset- SQL-like data filteringcheck_dataset_validity- Dataset validationget_dataset_parquet- Parquet file informationget_dataset_opt_in_out_urls- Opt-in/out URLsget_dataset_presidio_entities- PII entity detection
Development
Prerequisites
- Python 3.12+
- uv
Setup
make install # Install dependencies
make test # Run tests (38 tests)
make example # Test all tools
make run # Start server directlyDocker Commands
make docker-build # Build image
make docker-run # Run container
make docker-stop # Stop containerAPI Coverage
Implements all GET endpoints from the Hugging Face Dataset Viewer API:
| Endpoint | Tool | Description |
|---|---|---|
/splits | get_dataset_splits | Dataset splits information |
/info | get_dataset_info | Dataset metadata and features |
/first-rows | get_dataset_first_rows | Preview first 100 rows |
/rows | get_dataset_rows | Paginated data access |
/search | search_dataset | Full-text search within datasets |
/size | get_dataset_size | Dataset size information |
/statistics | get_dataset_statistics | Statistical analysis |
/filter | filter_dataset | SQL-like data filtering |
/is-valid | check_dataset_validity | Dataset validation |
/parquet | get_dataset_parquet | Parquet file information |
/opt-in-out-urls | get_dataset_opt_in_out_urls | Opt-in/out URLs |
/presidio-entities | get_dataset_presidio_entities | PII entity detection |
Built with FastMCP following all development best practices.
Docker
Build and Run
make docker-build
make docker-runWith Docker Compose
docker-compose up --buildDevelopment
TODO: Add development guidelines
API Documentation
TODO: Add API documentation
Contributing
TODO: Add contributing guidelines
License
TODO: Add license information
Quick Start
Claude Desktop Integration
Add to your Claude Desktop claude_desktop_config.json:
{
"mcpServers": {
"huggingface-mcp": {
"command": "docker",
"args": [
"run",
"--rm",
"-i",
"--name", "huggingface-mcp-claude",
"huggingface-mcp:latest"
]
}
}
}Build Docker Image
make docker-buildConfiguration
Copy the example environment file and configure as needed:
cp .env.example .env
# Edit .env with your configurationUsage
Running the Server
make runRunning Tests
make testRunning Examples
make exampleNo common issues documented yet. If you hit a problem, the repository's GitHub Issues page is the best place to look.