agent-browser

Warn

Audited by Socket on Feb 26, 2026

1 alert found:

Security
SecurityMEDIUM
SKILL.md

The fragment describes a legitimate and feature-rich browser automation tool intended for AI-agent workflows. While there is no explicit malicious payload, its powerful capabilities (state persistence, login/session reuse, network interception, and cross-environment support) introduce meaningful risk that must be mitigated with strong access controls, secret handling, auditing, and safe defaults. Overall, the content is BENIGN with MEDIUM risk factors due to potential data exposure, credential handling, and dependency/supply-chain considerations.

Confidence: 75%Severity: 75%
Audit Metadata
Analyzed At
Feb 26, 2026, 11:09 AM
Package URL
pkg:socket/skills-sh/AnEntrypoint%2Fplugforge%2Fagent-browser%2F@535cff36b226f971eb064c91fb3922597698ff47