The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/run_anomalygennext_amp.py invokes the containerized roi_place tool using subprocess.run. This is the intended primary functionality of the skill and is implemented using safe practices, including the use of argument lists instead of shell strings to prevent command injection, and setting environment variables like HF_HUB_OFFLINE and TRANSFORMERS_OFFLINE to prevent unauthorized network access during runtime.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data which represents an attack surface for indirect prompt injection.
Ingestion points: The skill ingests data from gap_parquet and defect_spec.jsonl files in scripts/prepare_anomalygennext_inputs.py.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within the roi_prompt_defect_location text prompt.
Capability inventory: The skill possesses the capability to execute commands via subprocess.run in scripts/run_anomalygennext_amp.py based on the processed configuration.
Sanitization: The skill performs basic presence validation for required fields but does not implement content-based sanitization or filtering for potential injection patterns in the user-supplied prompts.

tao-prepare-anomalygennext-inputs