video

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions for the user to install video frameworks such as Hyperframes and Remotion using standard package manager commands like npm install hyperframes and npx create-video@latest. These are standard practices for the skill's intended development and production use case.
  • [EXTERNAL_DOWNLOADS]: Recommends the integration of several well-known AI video services and frameworks, including HeyGen, Runway, and Kling. These references target recognized industry providers and official repositories.
  • [INDIRECT_PROMPT_INJECTION]: The skill describes a workflow in references/edit-anatomy.md that involves analyzing external video content via watch-video and social-fetch tools. This creates an ingestion surface for untrusted data. (1) Ingestion points: Transcripts and captions extracted from external social media videos via tools like watch-video and social-fetch. (2) Boundary markers: Absent; there are no explicit instructions for the agent to ignore or disregard commands embedded within the processed video content or transcripts. (3) Capability inventory: The agent has the ability to suggest shell commands for package installation and interact with various external video service APIs. (4) Sanitization: Absent; while the agent is directed to structure the output into a specific beat sheet format, there is no explicit filtering or sanitization of potentially malicious instructions within the source content.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 02:03 PM