ui-designer

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The prompt includes an explicit "CRITICAL OVERRIDE" that tells the agent to ignore base/system instructions and enforce TailwindCSS (i.e., to override higher-level system context), which is a concealed/overriding instruction outside the skill’s stated neutral design role — this is a prompt injection.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). This skill's Style Proposal Protocol and Phase 1 explicitly instruct the agent to "use search_web" and "Search for 3-5 competitor/reference designs via web search" and to include live reference sites from data/style-references.csv, meaning it fetches and interprets arbitrary public web pages as part of its decision workflow (SKILL.md), which could allow indirect prompt injection from untrusted third-party content.