The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides an explicit command to integrate a vendor-managed Model Context Protocol (MCP) server: claude mcp add --transport http ryux https://mcp.ryux.design/mcp. This is used to load specialized design reasoning tools into the agent's environment.\n- [EXTERNAL_DOWNLOADS]: The skill connects to an external infrastructure at https://mcp.ryux.design/mcp. This domain is owned by the vendor and is used to provide the skill's core analytical functionality, which is a standard use of vendor resources.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided UI/UX context, which represents an ingestion point for external data. To mitigate potential injection or quality issues, the skill defines strict 'Hard Gates' and 'Anti-Slop' protocols (e.g., RX-AS-01 through RX-AS-06) that require evidence-based decisions and explicitly instruct the agent to reject fabricated content or unfounded claims.

ryux-core