OpenMM MCP Server

MCP server that lets Claude configure, launch, monitor, and analyze OpenMM molecular dynamics simulations — including protein and membrane system setup, advanced sampling, and Abacus DFT calculations — through discrete tool calls.


Type	MCP server
Supplier	PhelanShao (community OSS)
Availability	GA — published on GitHub; stdio MCP server
Pricing	Free / OSS (GPLv3); OpenMM is MIT
Capabilities	Read/Write — writes simulation inputs and runs MD/DFT jobs on the host
Verified	works · 2026-07-20
Security	caution · 2026-07-20 — PhelanShao/openmm-mcp-server resolves, no OSV advisories, NOASSERTION license vs GPLv3 claim, single-maintainer + stale (2025-05-31)

How to install

This is a local stdio server. Clone it, install its Python dependencies, then register it.

git clone https://github.com/PhelanShao/openmm-mcp-server
cd openmm-mcp-server
pip install -r requirements.txt

(requirements.txt pulls in OpenMM and the task-management dependencies; a CUDA-capable GPU is needed for non-trivial production runs — OpenMM’s CPU platform works for small/tutorial systems.)

Claude Code — register over stdio (Claude launches the process itself; no separate terminal needed). Run this from inside the clone, or replace /path/to/openmm-mcp-server with the absolute path of your clone (e.g. /Users/you/repos/openmm-mcp-server, or $(pwd) if you are still inside it from the previous step):
```
claude mcp add --transport stdio openmm-server -- python /path/to/openmm-mcp-server/run_openmm_server.py
```
Claude Desktop — add the equivalent stdio entry to claude_desktop_config.json (set cwd to the absolute path of your clone):
```
{
  "mcpServers": {
    "openmm-server": {
      "command": "python",
      "args": ["run_openmm_server.py"],
      "cwd": "/path/to/openmm-mcp-server"
    }
  }
}
```
(The upstream README also shows an alwaysAllow array listing every tool name — that field is optional and auto-approves those tool calls; omit it if you want to confirm each invocation.)

What it does

Exposes discrete MCP tools for end-to-end MD and DFT work:

create_md_simulation / create_advanced_md_simulation — build a basic or advanced MD job (advanced integrators, barostats, constraints, metadynamics / free-energy sampling)
setup_protein_simulation — protein-system preparation template
setup_membrane_simulation — membrane-system preparation template
create_dft_calculation — DFT calculation via the Abacus engine
control_simulation — start / stop / pause running jobs
get_task_status / list_all_tasks — asynchronous task management with persistence
analyze_results — post-run analysis

Primary use cases: protein-stability MD, membrane-protein simulations, advanced free-energy/metadynamics sampling, quick DFT calculations driven from natural language.

Notes

Compute-heavy and write-capable. The server runs simulations on the host machine and writes outputs; production MD assumes a CUDA-capable GPU. Treat it as a long-running compute tool, not a read-only data lookup.
Licensed GPLv3 (the wrapper); OpenMM itself is MIT.
Complements the Molecular Dynamics (Claude Skill) — that skill teaches Claude to write and run OpenMM/MDAnalysis Python directly, whereas this server exposes MD/DFT as managed MCP tool calls with task tracking. Use PDB MCP Server / AlphaFold MCP Server for structure retrieval upstream and trajectory-analysis skills downstream.
Unverified — the exact requirements.txt contents and whether run_openmm_server.py is the canonical entry filename were read from the README excerpt only; confirm against the repo before a clean-room install.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.