Agent Browser
Audited by Socket on Feb 17, 2026
1 alert found:
Malware[Skill Scanner] Installation of third-party script detected This skill's documented capabilities are consistent with an automated browser CLI and do not contain explicit malicious code in the provided documentation. However, several legitimate features (eval <js>, upload, arbitrary navigation, snapshot --json) are high-risk channels for data access and exfiltration if misused or if the installed package/binaries are compromised. The installation via npm/git is typical but carries standard supply-chain risk; the documentation lacks artifact verification guidance. Overall: no direct malicious indicators present in this text, but the feature set requires careful trust and operational controls when used in untrusted environments. LLM verification: The SKILL.md describes a legitimate browser automation CLI whose features explain the presence of high-privilege operations (eval, cookies/storage access, file uploads, installer downloads). I found no direct evidence of malicious code or obfuscation in the supplied document. However, the feature set and installer flow produce a moderate supply-chain and data-exfiltration risk if the package source or install artifacts are untrusted. Recommended actions: install only from verified, pinned source