agent-browser

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill contains explicit commands that embed plaintext credentials (e.g., agent-browser fill @e2 "password123") and instruct filling form fields with username/password, which would require the model to output secret values verbatim.

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md outlines the agent’s runtime flow that performs agent-browser open <url> followed by agent-browser snapshot -i, which ingests page-derived text/structure from whatever URL is provided (outsider-authored web content), creating indirect prompt-injection risk via the snapshot/element text.