The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data for diagnosis, creating a surface for potential indirect prompt injection attacks through processed logs and error messages.
Ingestion points: Processes untrusted diagnostic data including RuntimeError messages, Kernel trace logs, and symbolic SM_ERR_* status codes extracted from user-provided failures in the environment.
Boundary markers: No specific boundary markers or instructions to ignore embedded content are provided for the error log processing steps.
Capability inventory: Possesses capabilities to read local documentation, inspect source code, and execute local validation tools like suggest_tests.py.
Sanitization: No explicit sanitization or filtering of the diagnostic content is specified before analysis.
[COMMAND_EXECUTION]: The skill instructs the agent to run local development and validation scripts, which involves executing shell commands.
Evidence: Recommends running python3 tools/agent/suggest_tests.py --paths <changed paths...> for validation and using the usd-brep-test-selection utility. These are standard operations for a developer-oriented skill within the project's own repository structure.

smlib-kernel-operations