multi-agent-runtime-engineering
Audited by Socket on Sep 19, 2026
2 alerts found:
SecurityAnomalySUSPICIOUS. The examples are internally consistent with the skill's stated purpose and do not show credential theft, hidden exfiltration, or untrusted download/execute behavior. However, the purpose itself is offensive multi-agent exploit development and pentest coordination, so this skill materially enables security testing/attack workflows and carries elevated operational risk even without malware indicators.
The fragment appears to be a legitimate multi-agent security-assessment coordination toolkit and documentation set, not an intentionally malicious payload. It contains potentially dangerous dual-use capabilities because it can launch intrusion, privilege-escalation, and lateral-movement agents against configured targets. Unsafe shell interpolation creates a meaningful command-injection and path-handling risk if inputs are attacker-controlled. No credential theft, exfiltration, persistence, or obfuscated malware is visible.