The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires shell permissions to execute the mediakit-cli binary for processing video data.
[EXTERNAL_DOWNLOADS]: The documentation instructs the user to install the @volcengine/mediakit-cli package from the npm registry. This is a vendor-provided dependency from volcengine required for the skill to function.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of untrusted external media files to extract text (OCR, ASR), which creates an attack surface for indirect prompt injection.
Ingestion points: Processes external video and audio files via video_url and audio_url parameters in tools such as video-ocr, asr-subtitles, and analyze-video-highlights (SKILL.md, reference/video-ocr.md).
Boundary markers: There are no explicit instructions or delimiters mentioned to separate extracted text from the agent's internal reasoning or system instructions.
Capability inventory: The skill possesses the shell permission to execute system commands through the CLI tool.
Sanitization: The skill does not describe any mechanisms for filtering, escaping, or sanitizing the text extracted from videos before it is incorporated into the agent's context.

byted-mediakit-video