alibabacloud-agentloop-management
Audited by Socket on Sep 23, 2026
2 alerts found:
SecurityAnomalySUSPICIOUS: the overall skill is largely coherent with Alibaba Cloud AgentLoop administration and mostly uses official Alibaba APIs/tools, but it includes risky supply-chain patterns. The main concern is credential forwarding (`ARMS_LICENSE_KEY`) into externally installed agent/plugin scripts whose provenance is not established in the skill, plus mutable shell installers and transitive skill installation. Without those installer paths it would be closer to benign documentation/workflow guidance.
The fragment documents a legitimate but highly privileged live-debugging and observability feature. It contains no direct evidence of malware or sabotage. Its security risk is primarily potential sensitive-data exposure and unsafe runtime code/expression execution if the service lacks strong authentication, authorization, sandboxing, redaction, and audit controls. Review the underlying implementation and API access controls before deployment.