Skip to main content
Glama

Mol-MCP: Computational Chemistry Interfaces for LLMs

WARNING

This project is a work in progress!

MolMCP is a modular set of interfaces for chemistry workflows. It provides a flexible architecture for building, configuring, and running agents and tools for chemical operatations.

Features

  • Modular agent and tool system

  • Extensible via configuration and plugins

  • Designed for quantum chemistry and molecular computation

Related MCP server: CovAiLent

Project Structure

src/
  molmcp/           # Main package
    agents/         # Agent servers
    tools/          # Tool servers
tests/              # Test suite
pyproject.toml      # Project metadata and dependencies
README.md           # Project documentation

Usage

Tools

uvx --from git+https://github.com/lukasmki/mol-mcp mol-mcp serve --help

Chat TUI

To use the built-in interactive agent with mol-mcp tools,

uv run mol-mcp go

Currently, the interactive agent is setup to use the Anthropic API. If you want to use another, take a look at the src/molmcp/agents/interactive.py script for an example of using fast-agent.

Installation

Install uv

# macOS
brew install uv

# Linux/WSL2  
curl -LsSf https://astral.sh/uv/install.sh | sh

The calc and geometry servers depend on tblite and require a Fortran compiler to complete installation.

# macOS
brew install gfortran

# Linux/WSL2
sudo apt install gfortran

In your project directory, install with uv

uv add git+https://github.com/lukasmki/mol-mcp

or

uv pip install git+https://github.com/lukasmki/mol-mcp

You can also run it as a uv tool outside of a project!

uvx --from git+https://github.com/lukasmki/mol-mcp mol-mcp serve --help

or install it to use it anywhere

uv tool install --from git+https://github.com/lukasmki/mol-mcp
mol-mcp serve --help

Easy Install with uvx: MCP-JSON Supporting Clients

Add the server into your MCP config:

{
    "mcpServers": {
        "MolMCP": {
            "command": "uvx",
            "args": [
                "-from",
                "git+https://github.com/lukasmki/mol-mcp",
                "mol-mcp",
                "serve",
                "smiles",
                "geometry",
            ]
        }
    }
}

On Windows with WSL2,

{
    "mcpServers": {
        "MolMCP": {
            "command": "wsl",
            "args": [
                "--shell_type",
                "login",
                "uvx",
                "-from",
                "git+https://github.com/lukasmki/mol-mcp",
                "mol-mcp",
                "serve",
                "smiles",
                "geometry",
            ]
        }
    }
}

Development Install

Clone the repository.

{
    "mcpServers": {
        "MolMCP": {
            "command": "uv",
            "args": [
                "run",
                "--project",
                "/path/to/mol-mcp",
                "mol-mcp",
                "serve",
                "smiles",
                "geometry",
            ]
        }
    }
}

On Windows with WSL2,

{
    "mcpServers": {
        "MolMCP": {
            "command": "wsl",
            "args": [
                "--shell_type",
                "login",
                "uv",
                "run",
                "--project",
                "/path/to/mol-mcp",
                "mol-mcp"
                "serve",
                "smiles",
                "geometry",
            ]
        }
    }
}

Contributing

Contributions are welcome! Please open issues or submit pull requests for bug fixes, new features, or improvements.

License

This project is licensed under the MIT License.

Available Tools

9 tools
geometry_buildC

Generates a geometry for the input SMILES

Inputs: format (required), smiles (required), output_file (required). Returns: message, success

geometry_optimizeC

Optimize the input geometry using xTB

Inputs: input_file (required), output_file (required). Returns: message, success

smiles_has_substructureA

Check if SMILES contains SMARTS pattern substructure

Inputs: mol (required), target_smarts (required). Returns: result

smiles_infoC

Return JSON formatted descriptors for a SMILES

Inputs: smiles (required). Returns: result

smiles_max_common_substructureC

Find maximum common substructure between SMILES

Inputs: mols (required). Returns: object

smiles_reactC

Apply a reaction SMARTS to a SMILES

Inputs: mol (required), reaction_smarts (required). Returns: result

smiles_removeC

Remove SMARTS pattern from a SMILES

Inputs: mol (required), target_smarts (required). Returns: result

smiles_replaceC

Replace SMARTS pattern in a SMILES

Inputs: mol (required), target_smarts (required), replacement_smarts (required). Returns: result

smiles_retrieveA

Retrieves a SMILES string from any non-SMILES chemical identifier

Inputs: chemical_identifier (required). Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.3/5.0

    Scored across 9 tools

    Disambiguation5/5

    Each tool has a clearly distinct purpose: geometry tools are separate from SMILES manipulation tools, and within SMILES tools each operation (substructure check, info, common substructure, reaction, remove, replace, retrieve) is unique and well-defined.

    Naming Consistency5/5

    All tool names follow a consistent 'prefix_action' pattern in snake_case, with 'geometry_' for geometric operations and 'smiles_' for SMILES operations, making the naming predictable and clear.

    Tool Count5/5

    The 9 tools cover a focused domain of computational chemistry operations without being excessive or too sparse, each earning its place for typical molecular manipulation tasks.

    Completeness4/5

    The tool set covers essential SMILES manipulations (retrieval, substructure search, reaction, editing) and basic geometry handling, but lacks some common features like molecular property calculation or format conversion, leaving minor gaps.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      ChemMCP is an easy-to-use and extensible chemistry toolkit for LLMs and AI assistants, enabling molecular analysis, property prediction, and reaction synthesis tasks without domain-specific training.
      74
      Apache 2.0
    • A
      license
      Not graded
      quality
      D
      maintenance
      An MCP server for chemistry-focused tools, enabling LLM agents to perform molecule parsing, format conversion, property lookup, and other chemistry operations with explainable responses.
      Apache 2.0
    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables natural language interaction with computational chemistry tools including molecule building, electronic structure calculations, geometry optimization, vibrational spectroscopy, and molecular dynamics, powered by RDKit, PySCF, and ASE.
      MIT
    • F
      license
      Not graded
      quality
      D
      maintenance
      Provides in-process RDKit cheminformatics operations — descriptors, fingerprints, similarity and substructure search, reaction enumeration, standardization, and 3D conformer generation — for AI agents.
      -