cmux-browser

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM context via runtime browser content: the skill’s cmux browser <surface> snapshot / get text / get html / find outputs page DOM text (authored by the visited site or other third parties) into the agent’s context after open/navigate to an arbitrary URL.