pyHMT2D MCP
A 2D hydraulic modeling package with an MCP server and skills: an agent opens an SRH-2D or HEC-RAS project and runs simulations.
- 134 ★ on the repository
- Last commit
Extended access
Why this level
- The agent edits model input files and runs solver executables on your machine.
- Long simulations and calibration use many resources.
Install
- Works in
Manual install
pip install pyHMT2D[ai]Installation with the agent extra. The README option with cloning the repository adds examples.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent edits model input files and runs solver executables on your machine.
- Long simulations and calibration use many resources.
This is third-party code. Review the repository files before installing.
What it does
PyHMT2D automates the SRH-2D and HEC-RAS two-dimensional hydraulic models: it reads results into Numpy arrays, converts them to VTK, compares simulations and prepares terrain. The `[ai]` extra adds the `pyHMT2D.AI_Tools.mcp_server` MCP server. Its tools open a project, change Manning's n and boundary conditions, run a simulation, query results and export VTK for ParaView. The `.agents/skills` folder holds nine command skills: `/hmt-open`, `/hmt-status`, `/hmt-modify`, `/hmt-run`, `/hmt-results`, `/hmt-export`, `/hmt-convert`, `/hmt-calibrate` and `/hmt-monte-carlo`. Each calls MCP tools in the right order. Calibration and Monte Carlo analysis also run through the agent.
Who it is for. Hydraulic engineers and researchers who model floods in SRH-2D or HEC-RAS and want to steer runs in plain language.
Good fit when
- Batch calibration of Manning's n against observations runs through `/hmt-calibrate`.
- A simulation needs an uncertainty estimate by Monte Carlo.
- SRH-2D results are exported to VTK for viewing in ParaView.
- A HEC-RAS model is converted to SRH-2D format to compare solvers.
Not a fit when
- You are not on Windows: the package targets Windows and Linux support is limited.
- You need another hydraulic model: SRH-2D, HEC-RAS and the toy Backwater-1D are supported.
- SRH-2D and HEC-RAS are not installed, and without solvers no simulation runs.
Example request
Open the Muncie SRH-2D project, calibrate Manning's n against observation data and export the result to VTK.MCP
- Transport
- stdio
- Authentication
- not required
Limitations
The main platform is Windows, because SRH-2D and HEC-RAS 2D exist for it. Solvers are installed separately and are not bundled. The agent edits project input files and runs the solver, so work on a copy of the model: source data and earlier results can be overwritten. Installation goes from a cloned repository with `pip install -e .[ai]` or from PyPI. The last commit in the repository is from August 2026.
How to disable. Remove the MCP server with `claude mcp remove pyHMT2D` or delete the entry from the client configuration. The skills disappear with the repository folder; the package is removed with `pip uninstall pyHMT2D`.
FAQ
How do I connect pyHMT2D to Claude Code?
Install `pip install pyHMT2D[ai]` and run `claude mcp add pyHMT2D -- python -m pyHMT2D.AI_Tools.mcp_server`. For Codex use `codex mcp add` with the same arguments.
Where do the skills come from?
From the repository `.agents/skills` folder. Agents with the Agent Skills standard find them: Claude Code, Cursor, Codex, GitHub Copilot, Windsurf and Gemini CLI; open the cloned repository.
Which models are supported?
SRH-2D and HEC-RAS 2D. The third, Backwater-1D, is a toy model for demonstrations.
Is Windows required?
Mostly yes. The author tested Linux with limitations and only with Linux versions of the solvers.
What does the `/hmt-calibrate` skill do?
It calibrates Manning's n against observations by calling MCP tools in order. A separate `/hmt-monte-carlo` covers result spread.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server with 35 tools for EnergyPlus building energy models: loading IDF files, editing parameters, simulation and plots.
The official MathWorks kit that gives an agent MCP tools and knowledge to work with Simulink models
Monte Carlo plugin for Claude Code and Cursor: table checks before queries, data incident triage and monitor creation.
The official MathWorks kit that connects an AI agent to MATLAB through an MCP server and curated skills
Reads COMSOL, Abaqus, Fluent and other CAE files into structured JSON for agents, without launching the solver.
OpenStudio MCP
openstudio-mcp
MCP server for building energy modeling: an agent builds an OpenStudio model, runs EnergyPlus and reads the results.
README in short
The README describes the package as a toolkit for automating SRH-2D and HEC-RAS: reading and writing results, VTK, converting meshes and results to 3D for OpenFOAM, comparing simulations and the `hmt-cli` CLI. The AI section adds an MCP server and nine skills.