browser-use

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill includes commands that require embedding secrets verbatim on the command line (e.g., browser-use cloud login , browser-use input ... "password", browser-use cookies set ), which forces an LLM or agent to handle and output secret values directly.

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

• Third-party content exposure detected (low risk: 0.10). The browser-use workflow ingests outsider-authored free text from whatever URL the user/agent navigates to via browser-use open <url> (followed by state/get text/get html/eval), meaning the agent reads attacker-controlled page content once a navigation target is selected.