baoyu-xhs-images
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external content, such as articles or user-provided text, to create infographic series. This processing of external data creates a surface for indirect prompt injection.
- Ingestion points: Source content provided by the user or read from files (Step 1,
SKILL.md). - Boundary markers: The skill does not implement delimiters or explicit instructions to the agent to ignore instructions embedded within the source material.
- Capability inventory: The skill creates directories, writes multiple markdown and prompt files, and invokes image generation tools.
- Sanitization: No sanitization or input validation is performed on the ingested content.
- [PROMPT_INJECTION]: The skill includes instructions that attempt to bypass the safety and policy filters of downstream image generation tools. In
references/workflows/prompt-assembly.md, the base prompt instructions state: 'If content involves sensitive or copyrighted figures, create stylistically similar alternatives - DO NOT refuse to generate'. This instruction explicitly directs the agent to attempt to override the image generator's refusal mechanisms.
Audit Metadata