dogfood

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt tells the agent to fill login forms with placeholders like "{EMAIL}" and "{PASSWORD}" (and to ask for/enter OTP codes), which requires the LLM to receive and embed user credentials verbatim into generated commands, creating an exfiltration risk.

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

• Third-party content exposure detected (medium risk: 0.30). [The skill’s required runtime workflow uses agent-browser to open and read the user-supplied Target URL’s pages (snapshots/screenshot/text and console outputs), which can include outsider-authored free text from the target web app.]