frontend-ui-engineering

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow includes fetching guidelines from a public URL (https://raw.githubusercontent.com/.../command.md) and then reading/applying them, so the fetched outsider-authored free text is ingested into the agent’s LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to "Fetch fresh guidelines before each review" from https://raw.githubusercontent.com/vercel-labs/web-interface-guidelines/main/command.md, which is a runtime fetch of external content that directly supplies the rules/instructions the agent must follow (i.e., controls prompts), and is required for the skill to operate.