pfc-mcp

English | 简体中文

PyPI Downloads GitHub stars License: MIT Python 3.10+

pfc3d>model new ;now, with LLM.

pfc-mcp connects AI agents to ITASCA PFC — Itasca's discrete element method (DEM) code — through the Model Context Protocol. Browse documentation, run simulations, and execute code, all through natural conversation.

pfc3d>model solve ;LLM solves.

Tools (10)

5 documentation tools — browse and search PFC commands, Python API, and reference docs. No bridge required.

5 execution tools — interactive REPL, task submission, progress monitoring, interruption, and history. Requires bridge.

First-time Setup

Prerequisites

• ITASCA PFC 6.0, 7.0, or 9.0 installed
• uv installed (for uvx)

Agentic Setup (Recommended)

Copy this to your AI agent and let it self-configure:

Fetch and follow this bootstrap guide end-to-end:
https://raw.githubusercontent.com/yusong652/pfc-mcp/main/docs/agentic/pfc-mcp-bootstrap.md

Manual Setup

1. Register the MCP server in your client config:

{
  "mcpServers": {
    "pfc-mcp": {
      "command": "uvx",
      "args": ["pfc-mcp"]
    }
  }
}

2. Start the bridge from inside PFC:

Download addon.py, then use either of these two flows inside PFC:

• Copy the file contents into the PFC IPython console and run them
• Or download the file and execute it in PFC GUI

Verify

Restart your AI agent (Claude Code, Codex CLI, Gemini CLI, etc.) and ask it to call pfc_execute_code to verify the connection.

Daily Startup

Once first-time setup is done, each new PFC session only needs the bridge re-started — paste addon.py into PFC's IPython console and you're back online. The MCP client config persists.

Features

• Multi-version PFC support - command docs for PFC 6.0, 7.0, and 9.0 via the version parameter
• Hierarchical documentation browsing - agents navigate the PFC command tree to discover capabilities and boundaries, reducing hallucinated commands
• Enhanced plot documentation - plot items reference docs supplementing the official documentation
• Interactive REPL - rapid iteration before committing to full scripts; agents can quickly test and refine code
• Task lifecycle management - submit long-running simulations, monitor progress, interrupt running tasks, and browse task history
• Multi-client compatible - works with Claude Code, Codex CLI, Gemini CLI, GitHub Copilot CLI, OpenCode, toyoura-nagisa, and other MCP clients

Troubleshooting

See Troubleshooting in the bootstrap guide.

Development

See Developer Guide: Install and Run from Source.

Contributing

PRs and issues are welcome! See the Developer Guide to get started.

License

MIT - see LICENSE.