UXarray MCP Server

An MCP server that lets an AI assistant (Claude Code, Claude Desktop, Cursor, or any MCP client) analyze unstructured climate meshes with UXarray — locally on your machine, or remotely on an HPC system you have access to.

┌─────────────┐  stdio  ┌──────────────┐                    ┌─────────────────┐
│  AI client  │ ◀─────▶ │ uxarray-mcp  │ ◀── Globus ──────▶ │  HPC endpoint   │
│  (Claude…)  │   pipe  │ (your laptop)│    Compute (opt)   │ (Slurm/PBS node)│
└─────────────┘         └──────────────┘                    └─────────────────┘

What the AI can do. Open meshes and datasets, compute area / zonal mean / vorticity / divergence, subset, remap, plot, and run multi-step workflows. All as natural-language prompts.

Local by default; HPC is opt-in. Everything runs on your machine unless you configure a Globus Compute endpoint. The remote option only becomes available once such an endpoint exists — running one requires an account and allocation on that HPC system, though a shared/service-account endpoint can let authorized users submit without their own login.

⚠️ What the AI can access. Any file you (or your HPC account) can read. Any compute the configured endpoint can submit. Outputs are written to your disk. See SECURITY.md before connecting any remote endpoint.

Pick your path

You are most likely one of:

1. Local user — laptop only, no HPC. → Local install.
2. HPC user, endpoint already exists — someone at your lab gave you a Globus Compute endpoint UUID. → Local install, then docs/remote-hpc.md.
3. HPC user, your own personal endpoint — you have a Globus identity and shell access to an HPC machine, and want to stand up an endpoint just for yourself. → Local install, then docs/operating-an-endpoint.md.
4. Group / shared endpoint operator — you're standing one up for a team, project, or lab. → Local install, then the full docs/operating-an-endpoint.md including service-account migration and the MEP allowlist.

Local install

Five steps. Each is one command unless noted.

Step 1 — Install the package

Pick one. uv is the easiest; pip works too.

# Recommended
uv tool install --python 3.12 uxarray-mcp

# Or from a fresh clone (developer path)
git clone https://github.com/UXARRAY/uxarray-mcp-server.git
cd uxarray-mcp-server && uv sync --python 3.12

Why --python 3.12? The server uses Globus Compute to submit work to HPC endpoints, and Globus Compute's serializer is fragile across Python minor versions — a 3.13 submitter against a 3.12 endpoint worker raises WorkerLost on non-trivial payloads. HPC sites broadly ship 3.12 conda stacks today, so we pin the install to match. Tracking removal of this pin at globus/globus-compute#2139. uv downloads 3.12 automatically if your system doesn't have it.

Step 2 — Write a starter config

uxarray-mcp setup

Creates ~/.config/uxarray-mcp/config.yaml with sensible defaults. Local mode needs nothing more.

Step 3 — Connect your AI client

Claude Desktop

uxarray-mcp install-claude        # merges the mcpServers block into your config
# or
uxarray-mcp install-claude --print-only   # prints the JSON to paste manually

Restart Claude Desktop. The uxarray server should appear in Settings → Developer.

Claude Code

claude mcp add uxarray --transport stdio -- uxarray-mcp serve

Then /mcp in Claude Code; pick uxarray.

Cursor / other MCP clients

Add an MCP server entry pointing at uxarray-mcp serve over stdio. See your client's MCP docs.

Step 4 — Sanity check

uxarray-mcp doctor

Should print local execution: ok and (if no endpoints configured) skip the remote checks.

Step 5 — Ask the AI to do something

In your client, try:

"Open <path to a UGRID/MPAS/SCRIP grid file> and plot the mesh."

That's it for local use.

Going beyond your laptop

If you have an HPC account at a national lab or university cluster with Globus Compute available:

Both paths assume you've finished local install above.

What the MCP exposes

Intent-shaped tools, not raw UXarray bindings:

• get_capabilities — what can I do with this mesh?
• analyze_dataset — deterministic first-look: inspect, validate, area, zonal mean, plots.
• run_analysis — one operation at a time (gradient, curl, subset, remap, …).
• plot_dataset — mesh, geographic, variable, or zonal-mean plots.
• diagnose_endpoint, probe_path_access — endpoint health + file readability.
• run_workflow, resume_workflow, get_status, get_result, manage_session — persisted sessions and multi-step workflows.

Tools that can run remotely take use_remote: bool and optional endpoint: str. The dispatcher falls back to local if the endpoint is unhealthy.

Full schema: docs/tools.md.

CLI reference

Risks (read before relying on output)

AI agents can misread prompts, pick the wrong file, get units wrong (e.g., sphere-radius scaling on derivatives), or run long jobs on your HPC allocation. uxarray-mcp does not guarantee correctness of agent-driven analysis. You are responsible for:

• Verifying numerical results before publishing.
• Reviewing what files the agent opens.
• Monitoring HPC job submissions against your allocation.

For the security model (what the agent and the endpoint operator can access), see SECURITY.md.

Development

uv sync --extra hpc --extra docs --dev
uv run pre-commit run --all-files
uv run pytest tests/ --ignore=tests/test_remote_agent.py
uv run sphinx-build -b html docs docs/_build/html

Release process: docs/release.md.

License

See LICENSE.