ai-image-editing
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill does not contain any malicious patterns, obfuscation, or unauthorized data access. It emphasizes ethical guidelines, such as the 'honesty spine,' which prohibits defect concealment, watermark removal, and unauthorized identity modifications. The workflow is designed around the 'TOUCH' framework, which prioritizes human judgment and legal compliance.
- [INDIRECT_PROMPT_INJECTION]: The skill processes images and metadata from external sources (e.g., photography, archives, and other generators). While this creates an attack surface for indirect prompt injection via embedded metadata or instructions, the risk is mitigated by the 'TOUCH' framework's 'Check the seams' pillar, which mandates 100% human verification of all results at full zoom before publication.
- [COMMAND_EXECUTION]: The skill mentions calling APIs for image editing engines (FLUX, Firefly) where connected, but these are legitimate integration points for its primary purpose. No arbitrary shell command execution or unauthorized process spawning was found.
Audit Metadata