Mozichem

MoziChem-MCP provides specialized MCP servers for chemical engineering and chemistry tasks, enabling AI assistants to perform thermodynamic calculations, equation of state modeling, and phase equilibrium analyses through a structured, protocol-driven interface.

How to use

To leverage the MoziChem-MCP servers with an MCP-compatible AI assistant, run the servers locally or access them through an MCP client. You can connect to the EOS models server to perform equation of state calculations, fugacity predictions, and phase behavior analyses, or run the flash calculations server for vapor–liquid equilibrium and multi-component phase calculations. Start each server independently and point your MCP client to the corresponding command and port. After starting, your AI assistant can issue high-level requests like calculating fugacities or performing flash calculations and receive structured results.

How to install

Prerequisites you need before installation are Python 3.13 or higher and the uv package manager (recommended). You can install from source or install from PyPI when available.

# Clone the project repository
git clone https://github.com/sinagilassi/mozichem-mcp.git
cd mozichem-mcp

# Install using uv (recommended)
uv sync

# Or install using pip (editable mode)
pip install -e .

Additional configuration and notes

Two MCP servers are provided for you to run independently: an EOS Models MCP Server and a Flash Calculations MCP Server. You can start either server using the uvx runner or directly via Python. The configurations below show the dedicated commands you can run to start each server.

# EOS Models MCP Server (uvx)
uvx --from mozichem-mcp mozichem-mcp-eos-models

# EOS Models MCP Server (Python, local run)
python -m mozichem_mcp.mcp.eos_models

# Flash Calculations MCP Server (uvx)
uvx --from mozichem-mcp mozichem-mcp-flash-calculation

# Flash Calculations MCP Server (Python, local run)
python -m mozichem_mcp.mcp.flash_calculation

Configuration for AI assistants

Configure your MCP client to expose both servers so your AI assistant can query EOS model calculations and flash calculations. The following example shows how to register both servers in a client that supports MCP configuration. You can use either the uvx-based commands or the Python module invocations shown above.

{
  "mcpServers": {
    "mozichem_eos": {
      "command": "uvx",
      "args": ["--from", "mozichem-mcp", "mozichem-mcp-eos-models"]
    },
    "mozichem_eos_py": {
      "command": "python",
      "args": ["-m", "mozichem_mcp.mcp.eos_models"]
    },
    "mozichem_flash": {
      "command": "uvx",
      "args": ["--from", "mozichem-mcp", "mozichem-mcp-flash-calculation"]
    },
    "mozichem_flash_py": {
      "command": "python",
      "args": ["-m", "mozichem_mcp.mcp.flash_calculation"]
    }
  }
}

Available tools

eos_models

EOS Models toolset provides equation of state calculations using Peng-Robinson, Soave-Redlich-Kwong, Redlich-Kwong, and van der Waals models, plus fugacity for pure components and mixtures and thermodynamic property predictions.

fugacity_calcs

Fugacity calculations for pure components and mixtures to support phase behavior analysis.

phase_behavior

Phase behavior analysis tools for predicting phase splits and phase envelopes under given temperature, pressure, and composition.

flash_calculation

Flash calculation tools for vapor-liquid equilibria, including multi-component and iterative convergence for bubble point and dew point conditions.