Skip to content
Tresca

OPM Flow MCP servers and AI agent skills

One community MCP server, opm-mcp, runs OPM Flow on ECLIPSE-format decks, reads the binary results and drives ResInsight, so an agent can inspect a deck, launch a run, check convergence and compare cases. It runs the Linux OPM binaries directly or, on Windows, through WSL2 with Ubuntu 24.04. The OPM project itself publishes no MCP server or skill.

  • No official Open source server yet.
  • 1 MCP server listed.
Open source1 serverChecked 4 Nov 2026

Integrations for OPM Flow

Community and commercial servers

MCP serverCommunity

opm-mcpby ojaogezi

Runs OPM Flow decks as background jobs, reads ECLIPSE-format results and drives ResInsight, using WSL2 on Windows.

LocalWindows, LinuxCan modify models4· updated 3 months ago

Listings are sorted by origin, then by recent activity. GitHub data is refreshed nightly. Nothing here is ranked.

How AI connects to OPM Flow

An MCP server or skill can only do what OPM Flow lets programs do. OPM Flow exposes 2 automation interfaces:

  • Python bindings (opm.common and opm.simulators) (Python): The OPM Python documentation covers running Flow from Python and the class documentation for opm.common and opm.simulators. opm.common parses decks and reads result files, which is how opm-mcp inspects keywords, wells and schedule steps before a run. The same page documents the PYACTION keyword, which calls a Python script from inside a running simulation to change controls at report steps. Documentation
  • ECLIPSE-compatible input decks (Input files): Flow reads the familiar DATA deck with its RUNSPEC, GRID, PROPS, SOLUTION and SCHEDULE sections and INCLUDE files, and writes UNSMRY, UNRST, EGRID and INIT results in ECLIPSE format. Because everything is plain text in and standard binaries out, an agent can clone a deck, change a keyword and rerun without any GUI. The OPM Flow manual is published under CC BY-SA 4.0, with the PDF on the GitHub release page. Documentation

What Open source offers

The Open Porous Media initiative publishes OPM Flow, ResInsight and opm-common under free licences, and its website and manual page mention no AI assistant or MCP server as of 4 November 2026. The opm-mcp server listed here is an independent community project that builds on the official Python bindings and Equinor's resdata reader.

Checked 4 Nov 2026. Sources: OPM project website, OPM Flow manual page

What you can do with OPM Flow and AI

  • Validate and summarise a DATA deck, including its INCLUDE tree, and list wells, phases and grid dimensions before you commit to a run.
  • Launch Flow as a queued background job, optionally with MPI, then poll its status and grep the log for shut-in wells or errors.
  • Pull field and well summary vectors, plot them in the chat, and run a material-balance and convergence check on the finished case.
  • Clone a deck with keyword overrides for a sensitivity or history-match step and compare the new run against the base case or observed data.
  • Export a ResInsight snapshot or an LAS well log for a well without opening the ResInsight GUI yourself.

Taken from the listings' own documentation. Check each listing for exactly what it supports.

Installing

The listings here document setup for Claude Code, Claude Desktop and Cursor. Install steps, copied from each project's own documentation, are on each listing's page. Most OPM Flow integrations run on your own computer next to OPM Flow, so you need OPM Flow installed and licensed on the same machine unless the listing says otherwise.

Questions

Is there an MCP server for OPM Flow?

Yes, a community one: opm-mcp by ojaogezi. It runs Flow, reads its results and drives ResInsight. The OPM project hasn't released its own server.

Can Claude run OPM Flow simulations on Windows?

With opm-mcp, yes. The server runs natively on Windows so Claude Desktop or Claude Code can start it, and it hands every OPM binary to an Ubuntu 24.04 distro under WSL2. Result files are parsed on the Windows side with resdata.

Does opm-mcp work with ECLIPSE decks?

It works with ECLIPSE-format DATA decks as far as OPM Flow supports their keywords, and reads the ECLIPSE-format result files Flow writes. Check unsupported keywords against the opm-simulators documentation before relying on a converted deck.

Where do PVT and relative-permeability tables come from?

opm-mcp deliberately leaves those to a pyResToolbox MCP server. Both pyResToolbox servers on Tresca generate PVTO, PVDG, SWOF and similar tables in ECLIPSE format.

AI-generated models, drawings and results must be checked by a qualified engineer before use.

Sources (checked 4 Nov 2026): OPM project website · OPM Flow manual page · OPM Python documentation · ojaogezi/opm-mcp README · OPM Flow product page

OPM Flow is developed by its own project (Open source). Tresca is independent and not affiliated with or endorsed by the OPM Flow project.