Skip to content
Tresca

CalculiX MCP (CAE Agent Hub)

MCP serverCommunity

Parses CalculiX .inp decks, edits design variables, runs ccx, reads .dat results and runs sizing optimisation on shells and beams. By Cai-aa.

1.3k starsUpdated 10 days agoGitHub data from 5 Nov 2026

At a glance

Type
MCP server
Works with
CalculiX
How it connects
Reads and writes files; the tool isn't needed
Access
Can modify models
Tool versions
unknown (uses the ccx executable on PATH or set in CCX_EXE)
Operating systems
Windows, macOS, Linux
Needs a CalculiX licence
No
Transport
Local (stdio)
AI clients documented
Not stated in its documentation
Licence
MIT
Language
Python

What it can do

  • Parses a deck for nodes, element counts, sections, materials and loads, and lists tunable variables such as shell thickness, beam section, material constants and load size.
  • Edits a single card in place by text replacement and writes a new .inp, so other cards are left untouched.
  • Judges a ccx run by stdout and the .sta file rather than the exit code, and computes von Mises stress, volume and mass itself from the .dat output.
  • Reads natural frequencies from *FREQUENCY steps and minimises mass under stress, displacement or frequency limits.

Limitations

  • Sizing optimisation works only on shell and beam models, not solids, and is not topology optimisation.
  • Requires mcp below 1.8 because it uses the older FastMCP import.

Install

uv venv .venv --python 3.11
uv pip install --python .venv/bin/python "mcp>=1.0,<1.8" meshio numpy "python-dotenv>=1,<2"

Source: github.com/Cai-aa/CAE-Agent-Hub/tree/main/MCP/CalculiX

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

Skills that come with it

SkillCommunity

CalculiX skills (CAE Agent Hub)by Cai-aa

Three SKILL.md workflows for the hub's CalculiX server: linear static runs, modal analysis and sizing optimisation.

1.3k· updated 10 days ago

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