diff

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill includes explicit instructions and code to bypass bot-management (UA/header spoofing, navigator.webdriver spoof, stealth headed escalation and challenge-solve loops), which is suspicious steering that weakens security boundaries even if intended for measurement/QA.

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/content-diff.mjs, the required runtime workflow navigates to outsider-supplied prototypeURL and buildURL and reads free text/DOM from those pages via page.evaluate(inventory, ...) (plus scrolling/overlays in scripts/live-session.mjs).