AEDT Evidence Electromagnetics
Use the repository toolbox and a user-owned AEDT installation. Never infer a
successful electromagnetic solve from ansysedt.exe, a project file, or a
geometry screenshot.
Start
ai-cae-toolbox context-scope --solver aedt
ai-cae-toolbox bridge-plan --solver aedt --objective "<objective>"
ai-cae-toolbox create-run --solver aedt --case <case-name> --objective "<objective>"
ai-cae-toolbox write-smoke-template --solver aedt --run-dir <run-dir>
If MCP is available, prefer:
aedt_check_installationaedt_run_pyaedt_script_filescan_run_evidencegenerate_run_report
Select the Correct Route
- Use direct PyAEDT when the requested solver or API surface is not exposed by the installed MCP.
- Use an AEDT MCP only for capabilities it actually exposes.
- Do not build HFSS when the task requires Maxwell, or claim that an HFSS-focused MCP completed a Maxwell workflow.
- Record AEDT version, PyAEDT version, project, design name, design type, solution type, and whether a new or existing desktop session was used.
Execute
Keep scripts in scripts/, source parameters in inputs/, project outputs in
outputs/, and field plots/tables in exports/. Make every geometry dimension,
material, excitation, boundary, setup, and sweep explicit.
For Maxwell matrix work, audit winding polarity and reciprocity. For HFSS, audit ports, modes, radiation boundaries, frequency setup, adaptive convergence, and S-parameter passivity. Read references/acceptance-gates.md before interpreting solver output.
Finish
Collect the AEDT project, solver/adaptive log, mesh statistics, numeric CSV/JSON, field images, and a report. Clearly label any external circuit calculation or Python-derived plot that is not a native AEDT result.
ai-cae-toolbox scan-evidence <run-dir>
ai-cae-toolbox generate-report <run-dir>