skills/phuetz/code-buddy/screenshot/Gen Agent Trust Hub

screenshot

Pass

Audited by Gen Agent Trust Hub on Mar 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides a set of terminal commands to interact with system utilities like scrot, screencapture, ffmpeg, and PowerShell for capturing visual data.
  • [DATA_EXFILTRATION]: Captures screenshots and recordings which could contain sensitive information. While no remote exfiltration is present, the skill enables the collection of sensitive visual data into local temporary files.
  • [PROMPT_INJECTION]: Suggestions to use OCR tools (tesseract) on screenshots create a vulnerability to indirect prompt injection, where instructions embedded in the visual content of the user's screen could manipulate the agent's behavior.
  • Ingestion points: Screen content processed via OCR (SKILL.md).
  • Boundary markers: None provided to delimit OCR-extracted text.
  • Capability inventory: Execution of system commands for screen capture and file management (SKILL.md).
  • Sanitization: None implemented or suggested for the data extracted from images.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 18, 2026, 07:13 AM
Security Audit — agent-trust-hub — screenshot