captions
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the transcription of untrusted video data, which could contain malicious instructions designed to influence the agent's behavior during subtitle editing.
- Ingestion points: Video files are processed using the transcribe and captions commands in SKILL.md.
- Boundary markers: No delimiters or instructions are used to isolate transcribed text from the agent's instruction context.
- Capability inventory: The skill uses subprocess calls via uv run to execute the video-agent utility for transcription and video processing.
- Sanitization: No sanitization is performed on the transcription output before it is potentially reviewed or edited by the agent.
- [EXTERNAL_DOWNLOADS]: The skill utilizes uv run to execute video-agent, which involves downloading the tool and its dependencies, such as mlx-whisper. The source appears to be associated with the skill author and well-known technology providers.
Audit Metadata