claude-photo-manager

Pass

Audited by Gen Agent Trust Hub on Feb 18, 2026

Risk Level: SAFE
Full Analysis
  • [Indirect Prompt Injection] (SAFE): The skill is designed to analyze untrusted visual content (screenshots, photos), which is a common surface for indirect prompt injection if images contain text instructions. However, as this is the primary intended purpose of the skill, it is considered a known capability risk rather than a malicious finding.
  • Ingestion points: scripts/analyze_screenshot.py reads image files provided via command-line arguments. SKILL.md documents processing of direct uploads and Base64-encoded strings.
  • Boundary markers: No specific delimiters are used to separate extracted visual data from system instructions in the provided scripts.
  • Capability inventory: The skill performs file reading (cv2.imread) and generates text-based reports on UI structure and colors.
  • Sanitization: The script focuses on structural image analysis (contours and color clustering) using OpenCV.
  • [EXTERNAL_DOWNLOADS] (SAFE): The skill documentation and scripts specify standard Python dependencies (Pillow, opencv-python) necessary for image manipulation. No downloads from untrusted or unverified remote sources are present.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 18, 2026, 01:34 PM