The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes python -m http.server 8888 & to host a local preview of the generated designs. This is a standard utility command for development workflows.- [EXTERNAL_DOWNLOADS]: The skill utilizes Gemini and Chrome MCP tools to retrieve external web data and screenshots for design analysis. These operations are performed through standard integrated tools.- [PROMPT_INJECTION]: The skill incorporates external data from URLs and research results into agent prompts. This presents a surface for indirect prompt injection (Category 8). Evidence Chain: 1. Ingestion points: $1 argument (route/URL), Gemini research output, and Chrome screenshots. 2. Boundary markers: Absent. 3. Capability inventory: Filesystem writes, Python HTTP server execution, and MCP agent spawning. 4. Sanitization: Not explicitly defined in the instructions. While the surface exists, no malicious patterns are present in the static skill definition.

design-catalog