video

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external content via tools like watch-video and social-fetch to analyze video structures, transcripts, and social media posts. This creates an attack surface where malicious instructions could be embedded in the external content (e.g., in video transcripts or visual frames) to influence the agent's behavior. The skill mitigates this by instructing the agent to present a 'beat sheet' for a 'review-once' human approval before any further actions are taken.
  • [COMMAND_EXECUTION]: The documentation includes instructions for the agent to execute shell commands such as npm install hyperframes and npx create-video@latest. These are standard project initialization and dependency management commands for the video frameworks described in the skill.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the download of third-party packages (Hyperframes and Remotion) from the npm registry. These downloads are associated with well-known and established programmatic video services mentioned throughout the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 01:11 AM
Security Audit — agent-trust-hub — video