AWS ECS and EKS now have remote MCP servers (aws.amazon.com)

AWS today put fully managed MCP (Model Context Protocol) servers for Amazon EKS and Amazon ECS into preview, giving AI assistants a standardized, cloud-hosted way to access live cluster context. Instead of running local MCP appliances, teams can use AWS-hosted servers that automatically handle updates and patching, integrate with AWS IAM for centralized access control, and record activity via CloudTrail—bringing enterprise-grade security, auditability and the usual AWS scalability and reliability to context-aware AI workflows. Technically, MCP supplies real-time cluster metadata and operational context to LLM-driven tools, improving the accuracy of code generation, debugging, and runbook guidance. The managed offering is designed to plug into popular developer assistants such as Kiro CLI, Cursor, and Cline to enable guided development, context-aware debugging, and operator-facing best-practice knowledge bases drawn from large-scale cluster experience. For the AI/ML community, this standardization fosters interoperable, safer assistant integrations with infra, lowers ops overhead for maintaining context services, and raises important considerations around IAM scoping, audit trails, and data access/latency. The feature is in preview; EKS and ECS documentation and launch posts provide setup and usage details.