browser-automation

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt includes examples that embed credentials and passwords verbatim in commands and JSON (e.g., agent-browser fill "password123", infsh interact with "password123", proxy_password "pass"), which requires the LLM to output secret values directly and poses an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.85). The required runtime workflow is browser navigation + snapshotting of the loaded page’s DOM to produce @e refs (i.e., it ingests outsider-authored free text from arbitrary URLs the user supplies via the “Navigate — Open a URL” step, which the LLM then uses as element text/attributes during snapshot/ref generation).

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill's Quick Start for the cloud provider runs an installation command that fetches and executes remote code ("curl -fsSL https://cli.inference.sh | sh"), which is a required dependency for the agentic-browser provider and therefore poses an execution risk.