openmc-validator-mcp
MCP serverCommunity
Pre-flight checks for OpenMC models over MCP: 14 tools that catch material, geometry, tally and data errors before a run. By zxkjack123.
0 starsUpdated 2 months agoGitHub data from 5 Nov 2026
At a glance
- Type
- MCP server
- Works with
- OpenMC
- How it connects
- Reads and writes files; the tool isn't needed
- Access
- Read-only
- Tool versions
- openmc-validator 0.4.0; Python 3.10 to 3.12
- Operating systems
- Not stated in its documentation
- Needs a OpenMC licence
- No
- Transport
- Local (stdio)
- AI clients documented
- VS Code (GitHub Copilot)
- Licence
- MIT
- Language
- Python
What it can do
- Validates materials, geometry, settings and tallies, flagging nuclide typos, unclosed surface references, bad batch settings and mis-ordered energy bins.
- Cross-checks requested nuclides and temperatures against cross_sections.xml, and tally filters against geometry cells and mesh extents.
- Converts an OpenMC energy tally to a FISPACT-II spectrum file and summarises statepoint.h5 tallies as JSON.
- Returns structured error codes with severity and a suggested fix for each finding.
Limitations
- Deeper checks, such as material round-trips and geometry-debug, need OpenMC installed from conda-forge; without it they degrade to warnings.
- The template and material catalogue tools need a separate openmc-templates checkout.
- Validates and post-processes only; it doesn't launch transport runs itself.
Install
pip install openmc-validatorSource: github.com/zxkjack123/openmc-validator-mcp#installation
{
"servers": {
"openmc-validator": {
"command": "openmc-validator",
"args": []
}
}
}Source: github.com/zxkjack123/openmc-validator-mcp#as-an-mcp-server
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.