image-crop

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • SAFE (SAFE): No malicious patterns, obfuscation, or unauthorized data access were detected. The skill defines a standard interface for image processing.
  • NO_CODE (SAFE): The skill does not include any scripts or executable files; it describes how to use an external CLI tool.
  • PROMPT_INJECTION (SAFE): The skill accepts input from URLs, creating a surface for indirect prompt injection. 1. Ingestion points: --in parameter in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Image cropping via agent-media command. 4. Sanitization: Assumed to be handled by the underlying media provider library (Sharp).
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 01:20 AM