screen-recorder

Warn

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill generates a custom Node.js script (record.mjs) and a Bash script (crop.sh) at runtime based on templates and executes them to perform browser automation and video processing.
  • [COMMAND_EXECUTION]: The skill relies on various system commands to verify the environment and process video files, including npx, ffmpeg, ffprobe, jq, and curl.
  • [EXTERNAL_DOWNLOADS]: During its preflight phase, the skill may prompt the user to download and install the Playwright package, the Chromium browser driver (approximately 150MB), and the ffmpeg utility.
  • [PROMPT_INJECTION]: The skill ingests untrusted data such as URLs, CSS selectors, and interaction sequences. These values are interpolated into the generated scripts, creating a potential surface for indirect prompt injection or logic bypass, although the skill uses JSON.stringify and action whitelists to mitigate these risks.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 20, 2026, 11:46 AM
Security Audit — agent-trust-hub — screen-recorder