video-editing
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted content (transcripts) to generate commands. Ingestion points: Video transcripts in Layer 2. Capability inventory: FFmpeg execution, ElevenLabs API, and Remotion rendering.
- [DYNAMIC_EXECUTION]: The workflow relies on generating and running shell scripts and code. Evidence: Scaffolding FFmpeg and Remotion code in Layer 2 and execution in Layer 3 and 4.
- [COMMAND_EXECUTION]: Uses FFmpeg via shell for video processing. Evidence: Multiple code blocks in Layer 3, 7, and 8.
- [EXTERNAL_DOWNLOADS]: Interacts with well-known services. Evidence: ElevenLabs API for voiceover generation and references to fal.ai and VideoDB.
Audit Metadata