web-design-guidelines

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly instructs fetching the latest guidelines from a public URL (https://raw.githubusercontent.com/vercel-labs/web-interface-guidelines/main/command.md in the "Guidelines Source" section), which the agent must read and apply to drive its review decisions, so arbitrary third‑party content could influence behavior.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly fetches the guidelines at runtime from https://raw.githubusercontent.com/vercel-labs/web-interface-guidelines/main/command.md and uses that fetched content to drive its rules and output format, meaning the remote file directly controls the agent's instructions and is a required runtime dependency.