agent-browser

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt includes examples and commands that embed passwords and credentials verbatim (e.g., agent-browser fill "password123", proxy URLs with user:pass, set credentials user pass), which would require the LLM to handle and output secret values directly.

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

• Third-party content exposure detected (high risk: 0.85). Outsider-authored free text is ingested when the workflow runs agent-browser open <url> followed by agent-browser snapshot -i / agent-browser get text body, which reads arbitrary page DOM/text from the runtime URL (public web content or other non-user-authored pages) into the agent’s LLM context.