baoyu-design-integration
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The documentation suggests installing the skill from an external GitHub repository (
JimLiu/baoyu-design) vianpxandgit clone. - [COMMAND_EXECUTION]: Instructions include running shell commands for local preview servers such as
python3 -m http.serverandnpx http-server. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted files like PDFs, screenshots, and Markdown, creating a potential for instructions in those files to hijack agent behavior.
- Ingestion points: Processes
brand-kit.pdf,outline.md,PRD.md, andscreenshot.png. - Boundary markers: Lacks instructions to the agent to treat external file content as untrusted or delimited data.
- Capability inventory: Writes HTML/JS files to local storage and executes shell commands for previewing outputs.
- Sanitization: No validation or sanitization of ingested content is documented.
- [DYNAMIC_EXECUTION]: The skill generates and saves HTML files with embedded JavaScript that are executed when viewed in a browser through the local preview server.
Audit Metadata