The Agent Skills Directory

[SAFE]: The skill defines a clear and legitimate workflow for video analysis, including task submission, polling for status, and handling timeouts. All operations are performed through dedicated tools (submitAideoTask, getAideoTaskStatus) appropriate for the skill's stated purpose.
[DATA_EXPOSURE]: The skill involves processing video data via external tools. This is a functional requirement for video analysis and does not involve unauthorized access to local sensitive files or credentials.
[PROMPT_INJECTION]: As the skill processes external video content (visual, audio, and dialogue), it is theoretically susceptible to indirect prompt injection where instructions could be embedded in the video stream. However, the skill acts as an analysis engine and does not have capabilities to execute arbitrary code or perform sensitive system operations that could be abused through such an injection.
Ingestion points: Video files/URLs passed through the multiInputs parameter in SKILL.md.
Boundary markers: None explicitly defined in the prompts.
Capability inventory: Limited to calling video analysis APIs and returning summaries or clip URLs.
Sanitization: None described, though the risk is minimized by the constrained output format.

analyzing-videos