browse-config

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt accepts proxy credentials (e.g., "http://user:pass@host:port" or username/password objects) and instructs the agent to write them into config files, which requires the LLM to handle and emit secret values verbatim (even if not shown to the user), creating exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.85). The skill explicitly guides creating "stealth" and "Google‑safe" browsing profiles for scraping protected sites and includes high-risk capabilities (proxy credential storage to disk, loading arbitrary addons/executables, disabling COOP, enabling mainWorldEval, env injection) that facilitate evasion, covert scraping, and the loading/execution of untrusted code—indicating clear potential for deliberate abusive use though it does not itself contain hidden exfiltration or an active backdoor.