The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted external content that is subsequently used to influence project modifications.
Ingestion points: External media files are ingested via import_media and download_media, and generated transcripts are read into the agent's context using find_transcript and read_captions (SKILL.md).
Boundary markers: The instructions lack explicit delimiters or guidance for the agent to treat transcribed text as potentially untrusted data rather than instructions.
Capability inventory: The skill utilizes capabilities like edit_captions, manage_transcript, and apply_script (which modifies timeline.md) to perform actions based on the processed content.
Sanitization: There is no mention of sanitizing or validating transcript data before it is interpolated into prompts or used to drive timeline edits.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch media from user-provided public URLs using the download_media tool when local assets are unreachable by the backend.

transcription