---
title: deepchem MCP Service
emoji: 🔧
colorFrom: blue
colorTo: indigo
sdk: docker
pinned: false
license: mit
---

# deepchem MCP Service

This deployment packages a FastMCP wrapper around the DeepChem Python repository and exposes practical tools for molecular featurization, dataset loading, split generation, and metrics.

## Available Tools

- `health_check`
- `list_modules`
- `list_symbols`
- `featurize_smiles`
- `split_indices`
- `classification_metrics`
- `load_molnet_dataset`

See detailed parameter and usage docs in `deepchem/mcp_output/README_MCP.md`.

## Local stdio (Claude Desktop / CLI)

```bash
cd deepchem/mcp_output
python start_mcp.py
```

Or run plugin entry directly:

```bash
cd deepchem/mcp_output/mcp_plugin
python main.py
```

## HTTP MCP (Docker / remote)

```bash
MCP_TRANSPORT=http MCP_PORT=8000 python deepchem/mcp_output/start_mcp.py
```

MCP endpoint:

- `http://localhost:8000/mcp`

## HuggingFace Spaces (Docker)

This package is configured for HF Spaces Docker runtime:

- service port: `7860`
- Docker entrypoint: `python deepchem/mcp_output/start_mcp.py`
- HTTP MCP endpoint: `/mcp`