multi-agent-collaboration

Warn

Audited by Socket on Sep 19, 2026

3 alerts found:

SecurityAnomalyx2
SecurityMEDIUM
SKILL.md

SUSPICIOUS. The skill is internally consistent with its stated purpose, but that purpose is to give an AI agent scalable offensive-security orchestration. Its multi-agent pentesting, exploitation workflow, and stealth/exfiltration content create high real-world risk even without hidden installers or credential theft paths in the file itself.

Confidence: 95%Severity: 89%
AnomalyLOW
payloads.md

This fragment primarily describes an authorized penetration-testing orchestration framework and supporting examples. It contains potentially dangerous active-scanning and exploitation capabilities, but the visible code does not demonstrate malicious supply-chain behavior, data exfiltration, persistence, or sabotage. The principal concrete security issue in the shown code is possible path traversal through unsanitized engagement_id values in CheckpointManager, along with weaker message replay and identity-validation design concerns. Missing implementations and call sites prevent complete assessment of actual runtime behavior.

Confidence: 96%Severity: 55%
AnomalyLOW
guides/task-decomposition-delegation-guide.md

The fragment appears to be a readable security-testing orchestration module. It contains no direct malware indicators or data-theft behavior, but it deliberately enables automated exploitation follow-up through sqlmap and RCE verification based on finding data. This creates a medium security risk if inputs or task execution are not restricted to authorized targets. The assessment is limited because the actual task executor and call sites are outside the shown fragment.

Confidence: 94%Severity: 58%
Audit Metadata
Analyzed At
Sep 19, 2026, 04:38 PM
Package URL
pkg:socket/skills-sh/brucesongs%2Fkali-claw%2Fmulti-agent-collaboration%2F@be2d5d64896ed8ae12a42c2268ba655ddb63ab3e88e6efba9a85510c19bd1529