editor
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on executing the
video-agentCLI tool viauv runfor various stages of the video production pipeline. This includes metadata extraction (info), transcription (transcribe), scene detection (detect), silence removal (tighten), and final assembly using Edit Decision Lists (edl).- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of transcribed audio from user-provided video files to determine the edit's narrative structure. - Ingestion points: Transcription output generated in Step 1 (
/tmp/t.txt) is read by the agent to determine the video's topic, structure, tone, and key lines. - Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are used when the agent reads the transcript data.
- Capability inventory: The skill can execute various video processing commands and write output files to the local file system.
- Sanitization: The transcript text is not sanitized or filtered before being analyzed by the agent to decide which beats and cuts to make.
Audit Metadata