agent-browser

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill includes commands and examples that require embedding secrets verbatim (passwords in fill commands, HTTP basic auth via "set credentials", proxy URLs with user:pass, cookie/localStorage setters, and header values), so the LLM would need to handle/output secret values directly.

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

• Third-party content exposure detected (high risk: 0.75). SKILL.md’s required workflow uses agent-browser open <url> followed by agent-browser snapshot -i, which reads the fetched page’s DOM-derived visible text/attributes (outsider-authored web content) and feeds that snapshot output (containing page text) into the agent/LLM context via the command results.