browser-qa
Audited by Socket on Sep 19, 2026
2 alerts found:
SecurityAnomalySUSPICIOUS/HIGH-RISK but not malware. The skill is internally coherent for browser-based security testing, yet it gives an AI agent offensive web-testing and anti-detection capabilities with broad execution tools. No direct credential theft or exfiltration is present, but the capability set is inherently risky and inconsistent with the "non-attack meta skill" label.
This is a security-testing and browser-QA payload collection, not apparent malware. It performs intrusive active testing and extensive sensitive-data collection, so it should only be run against authorized targets and with redaction, isolated credentials, and controlled evidence storage. The primary risks are operational misuse, credential exposure through logs/HAR/screenshots, and unintended impact from destructive or state-changing tests. No clear malicious exfiltration, persistence, command execution, or backdoor behavior is present in the supplied fragment.