Ask Claude what your Kubernetes and GPU fleet is really doing.
The ScaleOps connector gives platform engineers read-only access to live cluster data, so you can investigate workloads, resource consumption, GPU utilization, cost drivers, and automation history across every environment. Diagnose issues, spot rightsizing opportunities, and track capacity risks without leaving the conversation.
Or install manually from the Cursor Marketplace, or add this repository in Cursor settings.
The ScaleOps connector lets Claude users investigate and query the state of their Kubernetes clusters managed by ScaleOps.
- "What's the biggest cost optimization opportunity across our clusters this quarter?"
- "Show me workloads in production using less than 20% of their requested resources."
- "Which workloads in the payments namespace have the largest right-sizing opportunity?"
- "What CPU and memory does ScaleOps recommend for the api-gateway deployment?"
- "Were there any OOMKills or pod evictions in the staging cluster this week, and what does ScaleOps recommend?"
- "Which workloads are at risk of resource contention right now?"
- "Which workloads are currently in ScaleOps automation mode?"
- "What right-sizing changes did ScaleOps apply in the last 24 hours, and what was the impact?"
- "If we applied all pending ScaleOps recommendations, how many nodes could we remove from our EKS clusters?"
This plugin configures a single MCP server:
| Name | URL | Transport |
|---|---|---|
scaleops |
https://mcp.scaleops.com |
Streamable HTTP |
Requires an active ScaleOps account with the ScaleOps platform installed in your Kubernetes clusters. Authentication uses each user's existing ScaleOps SSO identity. Available to all ScaleOps customers.
MIT