The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection via external video content.
Ingestion points: The tools video_to_sfx and video_to_video_sfx accept a video_url parameter, allowing the ingestion of untrusted remote video files (SKILL.md).
Boundary markers: There are no explicit boundary markers or instructions to the agent to disregard potential metadata or embedded instructions within the ingested video content.
Capability inventory: The skill environment allows for Bash command execution, as well as Read and Write operations, which could be targeted by a successful injection.
Sanitization: Content processing is performed by the external Sonilo API backend; however, the agent's interaction with the resulting task data lacks specific sanitization constraints.

video-to-sfx