impeccable

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

• Third-party content exposure detected (high risk: 1.00). The skill's critique workflow explicitly directs the agent to open and inspect live pages by navigating to "direct URL" and injecting detection scripts (reference: reference/critique.md — "create a new tab and navigate to the page (use dev server URL for local files, or direct URL)"), and the brand guidance also directs sourcing public Unsplash images via public URLs (reference: reference/brand.md), so the agent is expected to fetch and interpret untrusted third‑party web content as part of its required workflow.

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

• Potentially malicious external URL detected (high risk: 0.80). The critique commands instruct running "npx impeccable live" (which will fetch and execute the "impeccable" package from the npm registry at runtime) and then injecting the served script (http://localhost:PORT/detect.js) into pages — this clearly fetches and executes remote code during runtime (flagged: npx impeccable / http://localhost:PORT/detect.js).