character-three-view
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user-supplied scripts and asset lists to derive prompts for image generation tools. This functionality presents a surface for indirect prompt injection where instructions embedded in the input data could attempt to influence the agent's actions or tool outputs. Evidence Chain: 1. Ingestion points: External script and asset list files. 2. Capability inventory: Invocations of image generation tools and local file system writes. 3. Boundary markers: Not explicitly defined. 4. Sanitization: No specific sanitization or filtering logic is mentioned for the ingested script content.
- [SAFE]: The skill utilizes modular references to other agent tools and configuration paths (e.g., cursor-image-gen and seedream-image-gen) to perform its core tasks. These references target standard skill directories and intended project folders, showing no signs of unauthorized access to sensitive system files or credentials.
Audit Metadata