agent-browser

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt shows embedding passwords directly in commands (e.g., fill @e2 "password123") and retrieving cookies/storage (which may contain session tokens), so the agent will be asked to include secret values verbatim in its outputs.

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly lets the agent "agent-browser open " and use "agent-browser snapshot -i", "get text/get html", "fill" and "click" to read and act on arbitrary external web pages (e.g., examples like https://example.com/products), so untrusted public web content can be ingested and materially influence subsequent actions.