The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's installation instructions direct users to download and install components from a third-party GitHub repository (JimLiu/baoyu-design) and through the NPM registry (npx skills add). These external sources are not part of the verified trusted vendor list.
[COMMAND_EXECUTION]: The skill requires executing multiple shell commands for setup and operation, including cloning repositories, using the npx package runner for installation, and starting a local web server using Python (python3 -m http.server) or Node.js (npx http-server). Running servers locally exposes the generated design directory to the local network.
[PROMPT_INJECTION]: The skill is designed to read and interpret data from external files—such as brand kits, project outlines, and application source code—to generate its design outputs. This behavior creates a surface for indirect prompt injection where malicious instructions embedded in those processed files could influence the agent's behavior.
Ingestion points: Accesses external files like brand-kit.pdf, outline.md, and existing codebase files (src/components/).
Boundary markers: The instructions do not define specific delimiters or security headers to differentiate between design instructions and the content of the ingested files.
Capability inventory: The skill performs file system writes to the designs/ directory and executes subprocesses to run a preview server.
Sanitization: There is no evidence of sanitization or validation logic to filter potentially malicious instructions within the data it processes.

baoyu-design-integration