web-files-image-handling

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides comprehensive and security-conscious patterns for client-side image processing. It includes specific guidance on preventing memory leaks via URL.revokeObjectURL, avoiding browser crashes by clamping canvas dimensions, and handling EXIF orientation correctly to avoid double-rotation.
  • [EXTERNAL_DOWNLOADS]: The skill references a metadata schema from the vendor's repository on GitHub (raw.githubusercontent.com/agents-inc/cli). This is a standard reference for metadata validation and traces to the skill's official vendor infrastructure.
  • [INDIRECT_PROMPT_INJECTION]: The provided patterns include ingestion points for user-supplied files (File and Blob objects in useImagePreview and useImageGallery). The patterns incorporate boundary markers such as MIME type validation (file.type.startsWith("image/")) and include sanitization logic like dimension constraints and size limits. The capabilities provided are limited to browser-side processing (Canvas API) with no network exfiltration paths or command execution triggers identified.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 02:48 PM
Security Audit — agent-trust-hub — web-files-image-handling