Skip to main content
Glama

demeter-mcp

MCP bridge for Demeter / Artemis / IFEFFIT EXAFS fitting. Lets Grok (or any MCP client) run headless fits, build Artemis projects, generate LaTeX reports, and launch the Artemis GUI.

Requirements

  • Windows (tested)

  • Demeter / Artemis (e.g. <DEMETER_INSTALL_DIR>)

  • Python 3.10+ with mcp package

  • XeLaTeX (optional, for PDF reports via demeter_fit_and_report)

Related MCP server: xraylarch-mcp

Grok configuration

Add to ~/.grok/config.toml:

[mcp_servers.demeter]
command = "D:\\anaconda3\\python.exe"
args = ["C:\\path\\to\\demeter-mcp\\server.py"]
env = { DEMETER_BASE = "D:\\DemeterPerl", DEMETER_PYTHON = "D:\\anaconda3\\python.exe" }
enabled = true
startup_timeout_sec = 30
tool_timeout_sec = 900

Restart Grok and verify with /mcps.

MCP tools (6)

Tool

Description

demeter_check

Verify Demeter/Perl/IFEFFIT

demeter_list_prj_records

List Athena .prj records

demeter_run_fit

Generic headless EXAFS fit

demeter_build_artemis_project

Build .fpj (2-shell a-MoO�?model)

demeter_fit_and_report

One-click Fit_XX batch (fpj + PDF + other/)

demeter_open_artemis

Launch Artemis GUI

One-click fit batch

demeter_fit_and_report with project_base pointing to your XAFS project root:

project_base = <YOUR_XAFS_PROJECT_DIR>

Defaults: data/MoOx.prj, results/Feff_aMoO3/feff.inp, ΔE₀ restrained to ±10 eV.

Output layout:

Fit_XX/
├── *.fpj
├── *_report.pdf
└── other/          # tex, json, logs

CLI (without MCP)

python scripts/fit_and_report.py config.json

License

MIT

Related MCP Connectors

Related MCP Servers

  • F
    license
    Not graded
    quality
    D
    maintenance
    Enables to interact with Abaqus FEA software through an MCP bridge, supporting connection checks, script execution, model queries, job submission, and simulation automation.
    3
    -
  • A
    license
    A
    quality
    A
    maintenance
    Enables controlling OriginLab Origin through MCP to read and write worksheets, build and style multi-panel graphs, and export figures, reporting what Origin actually did rather than what it was asked to do. Works from macOS or Linux against Origin running in a Windows VM.
    17
    29 PyPI
    3
    MIT