opm-mcp
MCP serverCommunity
Runs OPM Flow decks as background jobs, reads ECLIPSE-format results and drives ResInsight, using WSL2 on Windows. By ojaogezi.
4 starsUpdated 3 months agoGitHub data from 5 Nov 2026
At a glance
- Type
- MCP server
- Works with
- OPM Flow
- How it connects
- Drives a running copy of the tool
- Access
- Can modify models
- Tool versions
- OPM Flow and ResInsight from the Ubuntu 24.04 apt packages (tracks upstream releases); a custom Flow build can be set with OPM_MCP_FLOW_BIN
- Operating systems
- Windows, Linux
- Needs a OPM Flow licence
- No
- Transport
- Local (stdio)
- AI clients documented
- Claude Code, Claude Desktop, Cursor
- Licence
- MIT
- Language
- Python
What it can do
- Parses and summarises a DATA deck and its INCLUDE tree, lists wells and searches keywords across include files before a run.
- Starts Flow runs as queued background jobs (with optional MPI), tails the log and survives a server restart without losing track of running jobs.
- Reads UNSMRY, UNRST, EGRID, INIT and RFT files natively and returns summary vectors, field and well totals, material-balance checks and convergence reports.
- Writes modified copies of a deck for sensitivity or history-matching loops and compares the results against a base case or an observations CSV.
- Renders ResInsight snapshots and LAS well-log exports through ResInsight command files.
Limitations
- On Windows it needs WSL2 with an Ubuntu 24.04 distro, because the OPM binaries are Linux-only.
- Ships only a tiny SPE1 example deck; benchmark decks come from the separate OPM/opm-tests repository.
- Leaves PVT and relative-permeability calculations to a separate pyResToolbox MCP server rather than implementing them.
Install
git clone https://github.com/ojaogezi/opm-mcp.git
cd opm-mcp
py -m pip install -e .sudo apt-get update
sudo apt-get install flow resinsight python3-opm-common python3-opm-simulatorsclaude mcp add opm --scope user -- py -m opm_mcp.serverSource: github.com/ojaogezi/opm-mcp#register-with-claude-code
{
"mcpServers": {
"opm": {
"command": "py",
"args": ["-m", "opm_mcp.server"],
"env": {
"OPM_MCP_WSL_DISTRO": "Ubuntu-24.04"
}
}
}
}Source: github.com/ojaogezi/opm-mcp#register-with-claude-code
MCP servers and skills run code on your computer. Review the source before installing.
AI-generated models, drawings and results must be checked by a qualified engineer before use.