SU2 MCP Server

A Modular Compute Protocol (MCP) server that exposes SU2 workflows (session setup, configuration, solver execution, and results inspection) via JSON-schema tools. Each SU2 session operates inside an isolated working directory containing configuration, mesh, and outputs.

Features

• Session lifecycle management with per-session working directories.
• Configuration helpers for reading, parsing, and updating SU2 .cfg files.
• Mesh upload and automatic MESH_FILENAME synchronization.
• Safe solver execution wrappers with timeout and missing-binary handling.
• Utilities for inspecting output files, sampling surface data, and exporting results.
• Installation helper tools to verify SU2 availability.

Development

Environment

python -m venv .venv
source .venv/bin/activate
pip install -e .[dev]

Formatting, linting, and type checks

ruff check .
black .
mypy .

Tests

pytest

Running the server

The server is built with FastMCP to expose SU2 tooling with generated JSON Schema metadata. Install in editable mode to register the console script:

pip install -e .[dev]

StdIO (default)

Stream requests and responses over standard input/output with a single command:

su2-mcp-server --transport stdio

HTTP transports

Expose the MCP server over HTTP using either the standard HTTP transport (--transport http) or the Streamable HTTP variant (--transport streamable-http). Both respect host/port flags and path selection:

su2-mcp-server --transport http --host 0.0.0.0 --port 8002 --path /mcp
su2-mcp-server --transport streamable-http --host 0.0.0.0 --port 8002 --streamable-http-path /mcp

Server-Sent Events (SSE)

Expose the MCP server over HTTP with SSE. This starts a Starlette/uvicorn app on the configured host/port (defaults: 127.0.0.1:8000) with the SSE stream mounted at /sse:

su2-mcp-server --transport sse --host 0.0.0.0 --port 8000 --mount-path / --sse-path /sse

Additional options let you control the Starlette mount path (--mount-path) and message transport path (--message-path). If you need to embed the server programmatically, call su2_mcp_server.main.create_app or su2_mcp_server.fastmcp_server.build_server and invoke run on the resulting FastMCP instance.

Checking SU2 binaries

Use the get_su2_status MCP tool to report whether common SU2 binaries (e.g., SU2_CFD, SU2_DEF) are present on PATH. You can call it directly in Python after installing the package:

python - <<'PY'
from su2_mcp_server.tools.su2_installation import get_su2_status

print(get_su2_status())
PY

To install SU2 alongside the MCP server, opt in to the extra dependency group when installing:

pip install su2-mcp-server[su2]