conducting-wireless-network-penetration-test

Fail

Audited by Socket on Sep 14, 2026

2 alerts found:

SecurityMalware
SecurityMEDIUM
SKILL.md

SUSPICIOUS/HIGH-RISK skill. Its capabilities are internally consistent with a wireless pentest playbook, but the skill teaches an AI agent to perform offensive security actions: deauthentication, evil twin deployment, credential capture, password cracking, and internal network probing. There is no obvious malicious installer or third-party credential proxy, so this is not confirmed malware; however, the skill materially increases attack capability and supports direct credential theft techniques, making it high security risk.

Confidence: 95%Severity: 90%
MalwareHIGH
scripts/agent.py

High-confidence malicious/dual-use wireless attack tooling. The module includes functions to capture WPA handshakes (airodump-ng), check WPS (wash), crack captured handshakes (aircrack-ng), and send deauthentication frames (scapy.sendp). Although not all attack functions are invoked in main(), the presence of these capabilities makes the package unsuitable for untrusted installation and warrants immediate review/audit.

Confidence: 90%Severity: 85%
Audit Metadata
Analyzed At
Sep 14, 2026, 03:18 PM
Package URL
pkg:socket/skills-sh/mukul975%2Fanthropic-cybersecurity-skills%2Fconducting-wireless-network-penetration-test%2F@f8120a0cdfa22709e96ee466f41da84abe69d72ece05935ab1953db2ee5c4251