layer-reference-sets
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of user-provided images and metadata to train custom models and generate AI assets, creating a surface for indirect prompt injection.\n
- Ingestion points: The skill ingests user artwork and prompts via tools like
create_reference_setandexecute_forgeas described in SKILL.md.\n - Boundary markers: There are no explicit instructions or delimiters mentioned to isolate embedded instructions in user data from the agent's control flow.\n
- Capability inventory: The skill utilizes powerful tools (
execute_forge,start_training) that can perform high-cost operations and generate content based on untrusted inputs.\n - Sanitization: No input sanitization, filtering, or validation steps are defined for the user-supplied content.
Audit Metadata