gimi-illustration
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill ingest user-provided article text as part of its core functionality to create illustration prompts. This establishes an indirect prompt injection surface where instructions embedded in the input text could attempt to manipulate the image generation process. This risk is inherent to the skill's primary purpose.
- Ingestion points: User-provided article content processed in the main workflow (SKILL.md).
- Boundary markers: The workflow implements a logical separation between input and output through multiple steps (消化内容/出策略 then 组装Prompt), though it lacks explicit delimiters that instruct the model to ignore instructions within the input text.
- Capability inventory: The skill's capabilities are restricted to invoking platform-native image generation tools (image_gen or GenerateImage).
- Sanitization: The skill extracts specific nouns and metaphors from the text rather than directly interpolating raw input into the final tool call, providing a degree of implicit sanitization.
Audit Metadata