agent-browser

Warn

Audited by Socket on Apr 13, 2026

1 alert found:

Security
SecurityMEDIUM
SKILL.md

SUSPICIOUS: the capability set mostly matches a browser automation skill, and install sources are standard registries, but the footprint is high-risk for an AI agent. It can import existing browser auth, automate logins and other real-world actions, process untrusted web content with broad action authority, and optionally send data to a third-party AI gateway. This looks purpose-aligned but dangerous by default due to scope and data-handling breadth.

Confidence: 84%Severity: 74%
Audit Metadata
Analyzed At
Apr 13, 2026, 08:16 AM
Package URL
pkg:socket/skills-sh/vercel-labs%2Fagent-browser%2Fagent-browser%2F@67d654a553c1bd21b4cff0d9d61d75cd9a67e635