agkan-run
Warn
Audited by Socket on Apr 8, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The workflow is coherent for task-driven development, but its footprint is high-risk: it repeatedly performs external write actions (pushes and PRs), feeds untrusted task content into a powerful sub-agent, and depends on an unverifiable `agkan` CLI. This looks more like an aggressive automation skill than malware, but it is not low-risk.
Confidence: 88%Severity: 78%
Audit Metadata