offensive-zigbee-thread-matter

The skill is purpose-aligned but grants an AI agent offensive security assessment capabilities, which warrant high security risk under the supplied criteria. No evidence of credential theft, exfiltration, or confirmed malware is present.