The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation describes running tests using uv run, torch.distributed.run, and bash. These commands are used to execute local testing scripts and validation tools as part of a standard developer workflow for machine learning projects.
[EXTERNAL_DOWNLOADS]: The skill mentions automated downloading of missing test data from /opt/data and provides a utility script, download_golden_values.py, to fetch reference values from GitHub. These network operations are standard for maintaining test data parity across environments.
[PROMPT_INJECTION]: The skill describes an ingestion surface for indirect instructions via YAML recipes and JSON golden values. While this represents a theoretical attack surface common to testing frameworks, it is intended for processing local configuration files within the project repository and is considered safe for its primary purpose.

testing