ansys-mechanical-evidence-structural
GitHub指导Agent通过证据链完成Ansys Mechanical结构分析自动化,涵盖静力学、模态及非线性求解。支持RunWB2、ACT脚本或PyMechanical控制,强调模型审计、求解验证及多格式结果导出,确保分析过程可追溯且结论可靠。
触发场景
安装
npx skills add Cai-aa/CAE-Agent-Hub --skill ansys-mechanical-evidence-structural -g -y
SKILL.md
Frontmatter
{
"name": "ansys-mechanical-evidence-structural",
"description": "Evidence-first Ansys Mechanical structural automation through Workbench RunWB2 journals, Mechanical Scripting\/ACT ExtAPI, or PyMechanical gRPC. Use when Codex must build, solve, inspect, postprocess, or validate static, modal, thermal-structural, contact, nonlinear, or other Mechanical analyses with mesh metrics, MAPDL logs, reaction balance, RST files, numeric exports, and result images."
}
Ansys Mechanical Evidence Structural
Treat Mechanical as its own solver module. Workbench is an orchestration route, not proof that the Mechanical analysis solved.
Start
ai-cae-toolbox context-scope --solver ansys-mechanical
ai-cae-toolbox bridge-plan --solver ansys-mechanical --objective "<objective>"
ai-cae-toolbox create-run --solver ansys-mechanical --case <case-name> --objective "<objective>"
If MCP is available, prefer:
mechanical_check_installationmechanical_run_workbench_journal_filemechanical_run_python_controller_filemechanical_validate_result_summary_file
Choose One Control Route
- Use
RunWB2.exe -B -R <journal.wbjn>for project-level orchestration. - Use Mechanical Scripting/ACT with
ExtAPIfor model-tree operations. - Use PyMechanical gRPC for an external Python controller and explicit session lifecycle.
Record the selected route and do not let two controllers own the same session.
Model and Solve
Audit units, geometry scale, materials, contacts, joints, loads, constraints, large-deflection setting, solver controls, result requests, and mesh controls. Use stable names for bodies, named selections, contacts, loads, and results.
Before claiming success, read
references/acceptance-gates.md. A .mechdat,
Workbench project, or result PNG alone is insufficient.
Finish
Export .mechdat or .wbpj, file.rst, solve.out, mesh metrics, load and
reaction data, result CSV/JSON, PNG images, and report.md. State every
approximation such as bonded contacts, suppressed bodies, linear materials, or
disabled large deformation.
ai-cae-toolbox scan-evidence <run-dir>
ai-cae-toolbox generate-report <run-dir>
版本历史
- fc7b93e 当前 2026-08-16 08:57


