ui-design-critique

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow can ingest outsider-authored free text when it captures a live URL / running dev server and then extracts visible DOM text (e.g., via agent-browser/chrome-devtools during visual capture and detector evidence), which may include arbitrary third-party page content or user-generated text.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs running the external design.md tooling via npx (e.g., npx @google/design.md lint) and explicitly references the upstream repo at https://github.com/google-labs-code/design.md, meaning the agent may fetch and execute remote code from that package at runtime.