AgentChat-WebSubAgent

Warn

Audited by Socket on Oct 8, 2026

1 alert found:

Security
SecurityMEDIUM
SKILL.md

SUSPICIOUS. The skill's purpose matches multi-model orchestration, but it is over-broad: it mandates external transmission of local analysis for all tasks, uses a personal-repo script as the execution gateway, and relies on browser/web-session style provider access rather than narrow official APIs. This is not confirmed malware, but it poses meaningful supply-chain and data-leak risk.

Confidence: 86%Severity: 78%
Audit Metadata
Analyzed At
Oct 8, 2026, 07:42 AM
Package URL
pkg:socket/skills-sh/karlorz%2Fagentchat%2Fagentchat-websubagent%2F@ea683875f3f82b3bf7a4f03f5cc9ea38d078af4f