The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for executing local validation and rendering scripts through the uv package manager, which ensures a controlled environment.
Evidence: Validation instructions use uv run --locked --group gui python tools/scripts/smlib_gui_smoke.py to test workbench changes.
[DYNAMIC_EXECUTION]: The maintained tool is a developer workbench designed for interactive Python script execution and history management, facilitating a debugger-centric workflow.
Evidence: The skill defines the primary workbench purpose as managing "script execution/history" and preserving the reproducibility of UI-generated commands.
[INDIRECT_PROMPT_INJECTION]: The skill processes external geometric model files (e.g., USD, OCCT, SMLib), creating an ingestion surface for potentially untrusted data.
Ingestion points: Data is loaded via the shared dispatch logic in tools/smlib_gui/operations/file_io.py.
Boundary markers: Boundary markers for model parsing are not explicitly defined in the skill documentation.
Capability inventory: The skill uses isolated subprocesses for rendering and performs file system operations to save composite screenshots and render reports.
Sanitization: A strong mitigation is implemented through isolated model workers (--jobs N), which ensures that malformed inputs or worker failures do not corrupt the host process or affect other rendering tasks.

usd-brep-tools-python-gui