Skip to content
Tresca

PyFluent-MCP

MCP serverOfficial · Pre-1.0

Ansys's own MCP server for Fluent: launches or attaches to a Fluent session through PyFluent and runs sandboxed Python in it. By Ansys (Synopsys).

68 starsUpdated 1 day agoGitHub data from 5 Nov 2026

At a glance

Type
MCP server
Works with
Ansys Fluent / CFX
How it connects
Drives a running copy of the tool
Access
Can modify models
Tool versions
Any Fluent release that PyFluent (ansys-fluent-core 0.27 or later) can launch or attach to over gRPC
Operating systems
Not stated in its documentation
Needs a Ansys Fluent / CFX licence
Yes
Transport
Local (stdio), Remote (HTTP)
AI clients documented
VS Code (GitHub Copilot), Claude Desktop, Cursor
Licence
Apache-2.0
Language
Python
Release status
pre 1 0

What it can do

  • Launches a new Fluent session, attaches to a running one (local or remote) and disconnects on request.
  • Lets the agent browse the live Fluent settings tree, named objects and allowed values before it changes anything.
  • Runs or pre-checks Python in a persistent session behind an AST sandbox, rather than passing arbitrary code straight to the solver.
  • Summarises the setup, writes a simulation report, checks mesh quality, lists fields, compares case files and captures screenshots (20 tools in all).

Limitations

  • Version 0.6.0 on PyPI; the API is still pre-1.0.
  • Needs Python 3.12 or later, and every solver-facing tool needs a licensed local Fluent install.
  • Only find_api, get_help and validate_code work without Fluent; the HTTP transport is meant for trusted networks only, with no built-in authentication.

Install

pip install ansys-fluent-mcp
ansys-fluent-mcp

Source: github.com/ansys/pyfluent-mcp

MCP servers and skills run code on your computer. Review the source before installing.

AI-generated models, drawings and results must be checked by a qualified engineer before use.