Radar Kubernetes MCP Server llms.txt changelog

https://radarhq.io/llms.txt

− - [Benchmark hub](https://radarhq.io/benchmark): Three studies of AI agents diagnosing live Kubernetes faults from SREGym (Microsoft + UIUC) on a 3-node EKS cluster, one judge for every arm; each dataset is downloadable+ - [Benchmark hub](https://radarhq.io/benchmark): Three studies of AI agents diagnosing live Kubernetes faults from SREGym, an open academic benchmark, on a 3-node EKS cluster, one judge for every arm; each dataset is downloadable+ - [Radar Is #1 on SREGym: 95% of Live Kubernetes Faults Fixed](https://radarhq.io/blog/sregym-leaderboard): On SREGym's full 90-fault suite, Claude Code with Radar as its only cluster access diagnosed and fixed 95.2% of runs. The next entry scored 72.2%. Setup, misses, and how to rerun it.+ - [Radar at 4,000 Stars: Come Say Hello at KubeCon](https://radarhq.io/blog/radar-4000-github-stars): Radar passed 4,000 GitHub stars, and thousands of engineers now use it every day. Next week we get to meet some of them at KubeCon NA 2026, booth 365.

https://radarhq.io/llms.txt

+ - [Finding PHP2Shell in Kubernetes, Including the PHP Trivy Can't See](https://radarhq.io/blog/php2shell-kubernetes): PHP2Shell is fixed in PHP 8.2.35, 8.3.36, 8.4.27 and 8.5.12. How to find the Pods still running an older PHP with Radar, including images built on the official php image, where Trivy reports no PHP at all.+ - [New in Radar: Complete CloudNativePG Operations](https://radarhq.io/blog/cloudnativepg-ui): Radar now runs CloudNativePG day-2 work: which Postgres cluster needs attention, whether you could restore it, and switchover, fencing and restore that you review first.

https://radarhq.io/llms.txt

− - [Cluster audit](https://radarhq.io/product/cluster-audit): 36 best-practice checks across security, reliability, and efficiency− - [Kubernetes MCP server comparison](https://radarhq.io/kubernetes-mcp-comparison): Five servers ranked across diagnosis usefulness, speed, safety, writes, coverage, installation, multi-cluster support, and governance. Editorial order: 1. Radar, 2. containers/kubernetes-mcp-server, 3. Flux159/mcp-server-kubernetes, 4. k8sgpt, 5. Lens MCP. We build Radar; speed is measured separately and product claims link to primary sources+ - [Cluster audit](https://radarhq.io/product/cluster-audit): 44 best-practice checks across security, reliability, and efficiency+ - [Kubernetes MCP server comparison](https://radarhq.io/kubernetes-mcp-comparison): Six servers ranked across diagnosis usefulness, speed, safety, writes, coverage, installation, multi-cluster support, and governance. Editorial order: 1. Radar, 2. containers/kubernetes-mcp-server, 3. Flux159/mcp-server-kubernetes, 4. k8sgpt, 5. Amazon EKS MCP Server, 6. Lens MCP. We build Radar; speed is measured separately and product claims link to primary sources+ - [Kubernetes Rightsizing: Which Requests to Change, and When Not To](https://radarhq.io/blog/kubernetes-rightsizing-requests): How to right-size Kubernetes CPU and memory requests with Prometheus, and when HPA, OOMKills, bursty usage or throttling should stop a reduction.

https://radarhq.io/llms.txt

− - [Kubernetes MCP server benchmark](https://radarhq.io/benchmark/kubernetes-mcp-servers): Radar, containers/kubernetes-mcp-server, Flux159/mcp-server-kubernetes, k8sgpt, and raw kubectl on 25 live faults with the same agent and judge - across all faults, Radar averaged 57s to a diagnosis, 2.3x faster than the next-fastest MCP server; Radar still led at 47s on the 16 faults every arm diagnosed correctly; every scenario and score is downloadable+ - [Kubernetes MCP server benchmark](https://radarhq.io/benchmark/kubernetes-mcp-servers): Radar, containers/kubernetes-mcp-server, Flux159/mcp-server-kubernetes, the Amazon EKS MCP Server, k8sgpt, and raw kubectl on 25 live faults with the same agent and judge - across all faults, Radar averaged 57s to a diagnosis, 2.3x faster than the next-fastest MCP server; Radar still led at 33s on the 14 faults every arm diagnosed correctly; every scenario and score is downloadable+ - [Give Your Agent Context Before You Buy an AI SRE](https://radarhq.io/blog/ai-sre-give-your-agent-context-first): We ran the same agent and model on 50 live Kubernetes faults. With read-only cluster context it reached the answer about three times faster, and was slightly more accurate. Here's what it did without that context, and how to run the same test before you buy an AI SRE.