webm-reencode-for-chrome-seek

Warn

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: Dynamic Script Generation and Execution
  • The skill writes functional shell scripts to the temporary directory (/tmp/reencode-one.sh and /tmp/reencode-batch.sh) and executes them to perform video processing tasks.
  • These scripts utilize system tools such as ffmpeg, ffprobe, and stat on local video files.
  • [COMMAND_EXECUTION]: Command Injection Surface
  • The workflow uses the find command to collect file paths which are then passed as arguments to the generated shell scripts.
  • While the scripts use variable quoting, processing filenames from the filesystem without explicit sanitization provides an attack surface for command injection via maliciously crafted filenames.
  • [COMMAND_EXECUTION]: Environment Information Disclosure
  • The instructions contain hardcoded absolute paths (/Users/JWatkin70/Sites/magentaA11y/) that reveal the username and specific directory structure of the developer's workstation.
  • [COMMAND_EXECUTION]: Browser-Based Code Execution
  • The skill includes a Playwright-based verification step that executes custom JavaScript within a browser context to interact with a local development server.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 23, 2026, 04:50 AM
Security Audit — agent-trust-hub — webm-reencode-for-chrome-seek