browser

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill defines commands that take credentials and secret values as literal arguments (e.g., set credentials , cookies set , set headers ) and even examples filling a plaintext password, so an LLM producing those commands would need to insert secrets verbatim, creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly supports opening arbitrary URLs and reading page content (e.g., "agent-browser open ", "agent-browser get text/get html", "agent-browser snapshot" and the "爬取页面数据" and "打开网页并截图" workflows in SKILL.md), so it ingests untrusted public web content that the agent is expected to interpret and act on, enabling indirect prompt injection.