layer-image-editing
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill defines standard workflows for image editing using vendor-specific tools and models. The instructions focus on best practices for inpainting, outpainting, and upscaling without introducing security risks.
- [EXTERNAL_DOWNLOADS]: The skill provides instructions for installing sibling skills via the command 'npx skills add layerai/skills'. This targets the vendor's own resources and is considered standard practice for extending the agent's capabilities within the platform.
- [INDIRECT_PROMPT_INJECTION]: The skill describes an attack surface where the agent processes user-provided image editing instructions. Ingestion points: User instructions for image modification in SKILL.md. Capability inventory: Image processing tools like execute_forge and get_forge_run. Boundary markers: The documentation suggests using specific phrasing to keep subjects constant but does not define strict delimiters for user input. Sanitization: Not explicitly detailed in this documentation file.
Audit Metadata