browser-use

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt includes examples that embed secrets verbatim in commands (e.g., "password123" in a form input and '{"name":"token","value":"abc123"}' in a cookies set command) and advises passing API keys/cookies via CLI, which forces handling/output of secret values directly.

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly directs the agent to open arbitrary external websites (e.g., "uvx browser-use open https://news.ycombinator.com" and "open https://www.google.com") and to read/interpret page content via commands like "uvx browser-use state", "uvx browser-use extract", and "uvx browser-use eval", so untrusted public web content can influence clicks, inputs, and extraction logic and enable indirect prompt injection.