The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions to execute docker run commands for launching the TAO Toolkit container environment. It includes utility commands for directory creation (mkdir) and clearing build caches (find ... -exec rm). These operations are standard for the containerized workflow described and include security best practices such as running with non-root user IDs.
[EXTERNAL_DOWNLOADS]: The skill references and downloads container images from the NVIDIA Container Registry (nvcr.io). These are official vendor-owned resources from NVIDIA, the author of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external computer vision datasets consisting of image pairs and annotation files.
Ingestion points: External dataset files and annotation lists specified via the data_file spec parameter in actions such as train, evaluate, and inference.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content within these data files.
Capability inventory: The skill allows shell command execution via docker run and local file management via mkdir and rm.
Sanitization: The documentation does not specify any sanitization or validation logic for the content of the annotation files.

tao-train-fast-foundation-stereo