Usage examples

Check session status

Before you run CFX operations, ask the MCP client to call session_status. This shows whether a backend is connected and which capabilities are available.

Example prompt:

Check the current CFX MCP session status and tell me which actions are available.

Inspect model context

Use cfx_model_context for bounded inspection tasks, such as finding named objects, requesting API help, or retrieving current state snippets.

Example prompts:

Show a compact summary of the active CFX setup.

Find named boundary objects that contain inlet and return at most 10 matches.

Show API help and allowed values for the selected flow analysis path.

Run a workflow action

Use cfx_workflow for common CFX lifecycle tasks, including writing solver input, running a steady-state solver workflow, finding results files, and opening results in CFD-Post.

Example prompts:

Start CFX-Pre and import the mesh from C:/cases/mixer/mixer.gtm.

Write a solver input file to C:/cases/mixer/mixer.def.

Start the CFX solver for C:/cases/mixer/mixer.def, wait for completion, and report the result file.

Open the result file in CFD-Post.

Generate and validate code

Use codegen for small Python snippets and validate_code before running them through run_code.

Example prompt:

Generate a small PyCFX snippet to inspect the selected boundary condition. Validate it before execution.

Use generated code for custom model edits or detailed inspection. For standard lifecycle operations, use cfx_workflow so you receive typed results and consistent errors.

Handle errors

If a tool returns an error, inspect the message and retry with more context. Common recovery steps include checking session_status, narrowing your cfx_model_context query, or asking clarify for missing inputs before you run a workflow action.