molecule

Molecule-MCP acts as a bridge between molecular tools and Claude AI using the Model Context Protocol. It exposes dedicated MCP servers for PyMOL, ChimeraX, and a GROMACS Copilot server (optional) so Claude can interact with these tools as a co-scientist, performing tasks such as structure manipulation, visualization, and workflow automation within prompts. To leverage these capabilities, you configure the MCP integration inside Claude Desktop and ensure each tool server is accessible via the paths specified in the mcpServers configuration. Once set up, Claude can issue commands to launch and control PyMOL, ChimeraX, or the GROMACS Copilot server to carry out molecular modeling operations, analyses, or simulations in a prompt-driven workflow.

Prerequisites:

• Claude Desktop installed and running.
• Access to edit Claude Desktop configuration to include MCP servers.
• Python with pip disponible or your preferred environment to install mcp and dependencies.
• (Optional) Git to clone the molecule-mcp repository.

Installation steps:

1. Install the MCP package and the molecule-mcp tooling:
   • Run: pip install "mcp[cli]" chatmol
   • Run: pip install git+https://github.com/ChatMol/gromacs_copilot.git # optional for GROMACS Copilot
   • Verify: which mcp
2. Clone the Molecule-MCP repository and locate the scripts:
   • git clone https://github.com/ChatMol/molecule-mcp.git
   • cd molecule-mcp
   • pwd # note the path to molecule-mcp for later use
3. Configure Claude Desktop to recognize MCP servers:
   • Open Claude Desktop > Settings > Developer > Edit Config > claude_desktop_config.json
   • Add MCP server entries that point to your installed MCP runtime and the molecule-mcp server scripts, for example: { "mcpServers": { "pymol": { "command": "/path/to/mcp", "args": ["run", "/path/to/molecule-mcp/pymol_server.py"] }, "chimerax": { "command": "/path/to/mcp", "args": ["run", "/path/to/molecule-mcp/ChimeraX_server.py"] }, "gromacs_copilot": { "command": "/path/to/mcp", "args": ["run", "/path/to/molecule-mcp/mcp_server.py"] } } }
   • Replace "/path/to/mcp" and "/path/to/molecule-mcp/…" with the actual installed paths obtained from the previous steps.
4. Start or ensure Claude Desktop is running so the MCP configuration is loaded and the servers can be reached by Claude.

Notes:

• The PyMOL and ChimeraX server scripts are exposed under pymol_server.py and ChimeraX_server.py respectively; the GROMACS Copilot is exposed via mcp_server.py.

Tips and caveats:

• The configuration relies on absolute paths to the MCP runtime and the Molecule-MCP server scripts. Ensure these paths are correct to avoid connection errors.
• The GROMACS Copilot server is optional; include it only if you plan to use GROMACS-based Copilot automation.
• If Claude Desktop requires additional permissions, run Claude Desktop with appropriate access rights so it can invoke external scripts.
• Keep your Python environment and dependencies updated to avoid compatibility issues with the MCP framework.
• Validate each server individually by testing calls from Claude to ensure PyMOL, ChimeraX, and Copilot respond as expected.