humanize-image-generation
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and associated documentation focus exclusively on image generation workflows, art direction, and artifact auditing. No malicious patterns such as prompt injection, unauthorized data access, or malicious command execution were found.
- [COMMAND_EXECUTION]: The skill utilizes a local Python script (
scripts/compare_edit_regions.py) to perform raster image comparison. The script uses standard libraries (argparse,json,PIL) and follows safe practices for argument parsing and image processing. It is used locally to verify that non-target pixels remain unchanged during image editing sessions. - [EXTERNAL_DOWNLOADS]: The skill references several academic papers and official documentation from trusted sources (ArXiv, CVPR, Google Research, NIST, Adobe, Blender) to provide a research foundation for its methodologies. These are informative links and do not involve the execution of remote code.
Audit Metadata