baoyu-cover-image
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- Prompt Injection (LOW): The
base-prompt.mdfile contains the instruction 'DO NOT refuse to generate' regarding copyrighted or sensitive figures, which is an attempt to override the AI's safety guardrails. - Indirect Prompt Injection (LOW): The skill is susceptible to indirect prompt injection as it interpolates user-provided content into its image generation prompts. * Ingestion points: User-supplied content is used as input in
references/base-prompt.md. * Boundary markers: Uses triple dashes (---) and code block delimiters. * Capability inventory: The skill uses the 'nano banana pro' tool for creating images. * Sanitization: No sanitization or escaping of user input is performed.
Audit Metadata