The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use ffmpeg to generate contact sheets (ffmpeg -vf "fps=2") for visual inspection of the generated video output.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of subtitle text (passed via the subtitles parameter) and transcription hints (transcriptionPrompt). These inputs are interpolated into tool parameters, creating a surface where malicious instructions embedded in the data could attempt to influence the agent's behavior.
Ingestion points: subtitles input (Base64-encoded text), transcriptionPrompt field.
Boundary markers: No specific delimiters or safety instructions are defined for the untrusted text input.
Capability inventory: The skill utilizes model_run for video processing, ffmpeg for local file analysis, and asset_download for retrieving files.
Sanitization: The instructions mention Base64 encoding for data transport but do not specify content validation or sanitization routines.
[EXTERNAL_DOWNLOADS]: The documentation suggests installing additional tools using npx skills add scenario-labs/skills. This involves fetching and executing packages from the vendor's own repository.

scenario-caption-studio