Screenshot

Pass

Audited by Gen Agent Trust Hub on Jul 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute platform-specific commands for screen capture, including screencapture on macOS, scrot or grim on Linux, and nircmd on Windows. These are standard utilities for the intended functionality.
  • [EXTERNAL_DOWNLOADS]: The instructions suggest using npx playwright, which involves executing a package from the official npm registry. This is a standard practice for web automation tasks.
  • [PROMPT_INJECTION]: The skill facilitates the ingestion of external data by capturing screenshots of web pages (via Playwright or Puppeteer). This creates an attack surface for indirect prompt injection, where an attacker-controlled website could include malicious instructions meant to be interpreted by the AI agent when it processes the resulting image.
  • Ingestion points: URLs provided to Playwright/Puppeteer for screenshot capture in SKILL.md.
  • Boundary markers: None present; the skill does not explicitly instruct the agent to ignore instructions contained within the captured images.
  • Capability inventory: The skill has access to shell execution (screencapture, scrot, nircmd, etc.) and file writing (saving the screenshot output).
  • Sanitization: No specific sanitization or filtering of the visual content is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 28, 2026, 06:18 AM
Security Audit — agent-trust-hub — Screenshot