The Agent Skills Directory

[SAFE]: The skill provides standard documentation and guidance for the Ultralytics YOLO ecosystem. It correctly points to official vendor resources and follows legitimate usage patterns for vision AI tasks.
[EXTERNAL_DOWNLOADS]: Mentions installing the official ultralytics Python package from the public registry. This is required for the core functionality described in the skill.
[INDIRECT_PROMPT_INJECTION]: The skill involves processing external data such as datasets, images, and videos during model training and inference.
Ingestion points: Local or remote media files (source=video.mp4) and dataset configuration files (data=data.yaml).
Boundary markers: Not explicitly defined in the high-level instructions.
Capability inventory: Includes file system writes for model training outputs and network communication for streaming metrics to the Ultralytics Platform.
Sanitization: Performed by the internal parsing logic of the ultralytics library.

yolo