Tool Catalog

The current tool catalog is assembled by su2_mcp.fastmcp_server.build_server.

Available tool groups

• Session lifecycle: create_su2_session, close_su2_session, get_session_info

• Config helpers: get_config_text, parse_config, update_config_entries

• Mesh and setup: set_mesh, generate_mesh_from_step

• Solver execution: run_su2_solver, generate_deformed_mesh

• Results inspection: list_result_files, get_result_file_base64, read_history_csv, sample_surface_solution

• Availability and health: get_su2_status, ping

Tool bundle exports.

class su2_mcp.tools.PingRequest(*, message=None)

Request payload for the ping tool.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

model_config = {}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

class su2_mcp.tools.PingResponse(*, ok=True, message, server, timestamp)

Response payload for the ping tool.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

model_config = {}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

su2_mcp.tools.analyze_mesh(session_id)

Analyze the mesh attached to a session and return diagnostics.

Reports element counts by type, node count, boundary marker summary, and estimated solver runtime scaling factors to help diagnose latency.

su2_mcp.tools.close_su2_session(session_id, delete_workdir=False)

Close a session and optionally delete its working directory.

su2_mcp.tools.create_su2_session(base_name=None, initial_config=None, initial_mesh=None, mesh_file_name='mesh.su2')

Create a new SU2 session and return basic paths.

Without initial_config the session’s config.cfg starts from the project’s laptop preset: Euler, ROE with MUSCL, CFL 1 adaptive, 250 iterations, FARFIELD and WALL markers as the mesher names them, forces monitored on WALL, CL and CD in the history and on screen. The case is not set: MACH_NUMBER, AOA and REF_AREA must be set with su2_update_config_entries before su2_run_su2_solver will run. For the validated one-call route use su2_run_aero, which the aircraft-mcp gateway offers (this server alone does not).

su2_mcp.tools.generate_deformed_mesh(session_id, def_config_path=None, output_mesh_name='mesh_def.su2', max_runtime_seconds=600)

Run SU2_DEF to create a deformed mesh.

su2_mcp.tools.generate_mesh_from_step(session_id, step_base64=None, output_mesh_name='mesh.su2', geo_template_path=None, gmsh_timeout_seconds=600, surface_density=30, farfield_factor=10.0, surface_size_m=None, step_path=None)

Generate a 3D SU2 mesh from a STEP file and attach it to the given session.

Give the STEP either as step_path (the cad_path returned by tigl export_configuration_cad; preferred, the servers share a machine) or as step_base64 (the file content). Until 2026-10-08 only the content form existed, and a model-driven client that had to copy a 536,000- character string between two tools copied 528 of them.

By default this runs the same aircraft mesher the CPACS adapter and the paper’s runs use (Gmsh Python API: farfield box sized from the geometry, fragment against the imported solids, FARFIELD and WALL markers, a 3D algorithm fallback chain). It handles metre-scaled, multi-solid TiGL exports with intersecting wing and fuselage solids, which the static template cannot. Until 2026-09-30 the static millimetre-scaled template was the only path, and a colleague’s integration could not mesh the D150 through this endpoint at all.

Sizing matches the adapter: surface_density is the span-based preset knob (laptop 30, workstation 80, industry 200), and surface_size_m is the absolute chord-based cell size that takes precedence when given.

Pass geo_template_path to use a custom .geo template through the gmsh CLI instead (the template must Merge “model.step” and define FARFIELD and WALL physical surfaces).

Parameters:

• session_id – Existing SU2 session (create_su2_session first).

• step_base64 – Base64-encoded STEP file content (the cad_base64 field from tigl export_configuration_cad with include_base64=true). Not needed when step_path is given.

• output_mesh_name – Filename for the mesh in the session workdir.

• geo_template_path – Optional path to a .geo file (CLI path).

• gmsh_timeout_seconds – Timeout for the gmsh CLI subprocess.

• surface_density – Span / near-field cell size ratio (default 30).

• farfield_factor – Farfield box extent in spans (default 10).

• surface_size_m – Absolute near-field cell size in metres; overrides surface_density when set.

• step_path – Path of the STEP file on this machine (the cad_path from tigl export_configuration_cad). Preferred over step_base64.

Returns:

Dict with mesh_path, success, and optional error.

su2_mcp.tools.get_config_text(session_id)

Return raw config text for the session.

su2_mcp.tools.get_result_file_base64(session_id, relative_path, max_bytes=104857600)

Return a base64-encoded slice of a result file.

su2_mcp.tools.get_session_info(session_id)

Return session paths and last run metadata.

su2_mcp.tools.get_su2_status()

Return SU2 availability information for the host.

Returns:

Mapping that indicates whether any SU2 binaries are available alongside per-binary details.

Examples

>>> status = get_su2_status()
>>> set(status.keys()) == {"installed", "binaries", "missing"}
True

su2_mcp.tools.list_result_files(session_id, extensions=None)

List result files produced within a session directory.

su2_mcp.tools.parse_config(session_id)

Parse the session configuration into key/value entries.

su2_mcp.tools.ping(args)

Return a simple health check response without requiring SU2.

The tool intentionally avoids importing SU2 or touching the filesystem so that MCP clients can verify connectivity even when SU2 binaries are unavailable.

su2_mcp.tools.read_history_csv(session_id, relative_path, columns=None, max_rows=1000, skip_rows=0)

Read a subset of a history CSV file.

su2_mcp.tools.run_su2_solver(session_id, solver='SU2_CFD', config_override_path=None, max_runtime_seconds=600, capture_log_lines=100, allow_incomplete_config=False)

Run a SU2 solver process and capture output metadata.

For SU2_CFD the configuration must set MARKER_MONITORING, MACH_NUMBER, AOA and REF_AREA; otherwise the tool returns an error of type config_incomplete and runs nothing. Without MARKER_MONITORING SU2 evaluates the force coefficients on no surface and writes CL = CD = 0.0 (two model-driven runs on 2026-10-08); without the other three it runs a case nobody specified. allow_incomplete_config=True skips the check for runs that want no force coefficients.

su2_mcp.tools.sample_surface_solution(session_id, relative_path, marker_name, fields, max_points=5000)

Sample surface solution fields from a CSV-like file.

su2_mcp.tools.set_mesh(session_id, mesh_base64, mesh_file_name='mesh.su2', update_config=True)

Persist a mesh file for the session and update the config.

su2_mcp.tools.update_config_entries(session_id, updates, create_if_missing=True)

Update configuration entries for a session.