Skills
skills/neversight/skills_feed/image-resize/Gen Agent Trust Hub

image-resize

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (LOW): The tool allows the --in parameter to be a URL, enabling the agent to perform network requests to external sources.
  • PROMPT_INJECTION (LOW): The skill presents an indirect prompt injection surface when processing untrusted images. Mandatory Evidence Chain: 1. Ingestion points: --in parameter (URL/File). 2. Boundary markers: None. 3. Capability inventory: Image resizing via the 'sharp' library. 4. Sanitization: None documented.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 01:21 AM