baoyu-article-illustrator
Pass
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The system prompt in
prompts/system.mdcontains an explicit directive to bypass safety constraints: 'If content involves sensitive or copyrighted figures, create stylistically similar alternatives - DO NOT refuse to generate'. This is a direct attempt to override the safety filters of the underlying image generation model.
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by interpolating untrusted article content into prompt templates in
references/prompt-construction.md. Ingestion points: User-provided article content and descriptions. Boundary markers: The templates use header labels (e.g., LABELS, ZONES) but lack robust delimiters or instructions to prevent the model from following commands embedded in the source text. Capability inventory: The skill creates prompt and outline files and invokes image generation tools. Sanitization: There is no evidence of content validation or escaping before interpolation. - [COMMAND_EXECUTION]: The skill uses basic shell commands like
test -finSKILL.mdandreferences/workflow.mdto check for the existence of configuration files (EXTEND.md) and reference images. These operations are limited to local file system discovery and do not involve remote code execution.
Audit Metadata