groom-backlog-item
Warn
Audited by Socket on Mar 29, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The core behavior mostly matches backlog grooming, but the skill grants broad autonomous research, command execution, and GitHub write/close actions, and it treats externally supplied procedural content as executable instructions. The largest trust concern is the unverified plugin_dh_backlog MCP dependency, which is entrusted with local backlog writes and GitHub synchronization.
Confidence: 84%Severity: 74%
Audit Metadata