video-editing
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The workflow relies on generating and running FFmpeg commands and Remotion scripts. A risk exists if parameters derived from untrusted source material (like video transcripts) are directly interpolated into shell commands without escaping, potentially leading to command injection.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. It encourages the agent to analyze transcripts of raw footage to plan edits and generate code. An attacker could embed instructions within a video's audio/transcript to manipulate the agent's output during the organization phase.
- Ingestion points: Video transcripts and recordings processed in the 'Organization' layer (SKILL.md).
- Boundary markers: The instructions lack delimiters (e.g., XML tags or triple quotes) to separate untrusted transcript data from system instructions.
- Capability inventory: The agent has access to shell execution for FFmpeg and Remotion, as well as network access for asset generation.
- Sanitization: There is no explicit logic to validate or sanitize the data extracted from transcripts before it is used to scaffold commands.
Audit Metadata