baoyu-xhs-images
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from user-provided files or pasted text during the "Analyze Content" phase.
- Ingestion points: User-provided file paths or pasted content saved to
source.md(documented inSKILL.md). - Boundary markers: The instructions do not specify explicit delimiters or "ignore" warnings when reading the source content for analysis.
- Capability inventory: The skill performs file writes (analysis markdown, outlines, and prompts) and invokes image generation tools. It does not perform network operations to non-whitelisted domains or execute arbitrary system commands.
- Sanitization: No content sanitization or validation logic is defined for the input text before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill uses a slash-command syntax (
/baoyu-xhs-images) as a trigger for agent behavior. This is a design pattern for the agent's internal reasoning and does not involve the execution of shell commands on the host system. - [SAFE]: No obfuscation, persistence mechanisms, or unauthorized privilege escalations were detected. The image generation process in
references/base-prompt.mdincludes instructions to use cartoon alternatives for sensitive figures, which is a creative steering instruction rather than a malicious safety bypass.
Audit Metadata