agent-browser

SUSPICIOUS. The skill is coherent with its claimed purpose and uses an official install path, so it is not malicious by itself. However, its footprint is high-risk for an AI agent: it can log into arbitrary sites, persist auth state, upload files, route traffic through proxies/providers, ignore HTTPS errors, and execute browser-side JS, making it a powerful automation surface that requires strong user control.

No direct evidence of malware is present in the provided fragment because it is documentation rather than executable code. However, it describes a high-privilege browser automation interface that can execute arbitrary in-page JavaScript (including via base64/stdin), modify and persist sensitive browser auth/state, intercept/mock network traffic, load arbitrary extensions, and weaken transport security. The security concern is misuse/abuse potential and risk escalation in compromised or untrusted automation contexts.