word-highlight
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions are focused on image manipulation (cropping, thresholding, bounding box calculation) and drawing overlays for highlighting text. No malicious patterns, obfuscation, or data exfiltration attempts were detected.
- [DATA_EXPOSURE]: The skill operates on user-provided images and text strings. It does not attempt to access sensitive system files, environment variables, or hardcoded credentials.
- [REMOTE_CODE_EXECUTION]: There are no commands to download or execute remote scripts. The mention of 'video.js' and 'words.js' refers to local context files within the user's project workflow.
- [PROMPT_INJECTION]: The instructions do not contain attempts to override safety guidelines, ignore previous instructions, or extract system prompts.
Audit Metadata