xraylarch-mcp

MCP (Model Context Protocol) server that wraps xraylarch for X-ray spectroscopy analysis. Load, process, analyze, and plot XAS/XAFS spectra through Claude Desktop or Claude Code with natural language tool calls. Supports XANES normalization, EXAFS extraction, Fourier transforms, linear combination fitting, PCA, FEFFIT shell fitting, and publication-quality plotting.

Installation

pip install xraylarch-mcp

Requires Python 3.11 or newer. The server is built on version 2 of the MCP Python SDK (mcp>=2).

Related MCP server: fdmnes-mcp

Configuration

Claude Desktop

Add to your Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "xraylarch": {
      "command": "python",
      "args": ["-m", "xraylarch_mcp"]
    }
  }
}

Claude Code

claude mcp add xraylarch -- python -m xraylarch_mcp

Quick Start

Ask Claude:

"Load the spectrum at ~/data/fe_foil.xdi and show me the normalized XANES"

Claude will:

1. Call larch_load_spectrum to read your data

2. Call larch_normalize to perform pre-edge subtraction and normalization

3. Call larch_plot with plot_type="norm" to generate a publication-quality plot

Tool Reference

I/O Tools

XAFS Analysis Tools

Plot Tools

Math Tools

Fitting Tools

Interactive Tools

Interactive normalization workflow

1. Ask Claude to interactively normalize your spectrum

2. An HTML page opens in your browser with Plotly.js plots and sliders

3. Adjust E0, pre-edge range, post-edge range, and polynomial degree — plots update in real time

4. Click Save Parameters to download a JSON file, or Copy to Clipboard

5. Tell Claude to apply the saved parameters

"Load Pdfoil.prj and let me interactively choose the normalization parameters"

Advanced

Resources

The server exposes documentation resources that Claude can read on demand:

• larch://docs/xafs — XAFS analysis reference

• larch://docs/io — I/O functions reference

• larch://docs/fitting — Fitting reference

• larch://docs/feff — FEFF interface reference

• larch://docs/interactive — Interactive tools reference

• larch://docs/examples/xanes — XANES workflow example

• larch://docs/examples/exafs — EXAFS workflow example

Links

• xraylarch documentation

• MCP specification

• Source code

Releasing

Releases go to PyPI from GitHub Actions (.github/workflows/publish.yml), with no token: PyPI trusts this repository's workflow directly.

1. Set the new version in pyproject.toml and src/xraylarch_mcp/__init__.py, and merge that to main.

2. Tag the merge commit and push the tag, e.g. git tag v0.4.1 && git push origin v0.4.1, or publish a GitHub release with a new tag v0.4.1.

The workflow runs the tests, checks that the tag matches both version strings, builds the package and uploads it.

One-time setup: in the project's PyPI settings under Publishing, add a GitHub publisher with owner Joint-Photon-Sciences-Institute, repository xraylarch-mcp, workflow publish.yml and environment pypi.

License

BSD-3-Clause (matching xraylarch)