K8sGPT is an open-source tool that gives Kubernetes clusters an AI-assisted health check: it scans your cluster, finds problems, and explains them — and how to fix them — in plain English. It encodes the knowledge of experienced SREs into built-in analyzers and uses an LLM backend of your choice for the explanations. The project has about 8,300 stars on GitHub and is actively developed in 2026.
K8sGPT runs as a CLI or as an in-cluster operator. Its analyzers inspect common failure sources — pods, services, deployments, nodes, gateways and more — and surface issues such as crash-looping containers, missing resources or misconfigurations. The raw findings are then passed to an AI backend (OpenAI, or local and alternative providers) that translates them into readable, actionable explanations. Filters and anonymization options help teams use it safely, and integrations with tools like Trivy extend analysis to security scanning of workloads. In operator mode it can publish metrics for Prometheus, cache results remotely so repeated scans stay cheap, and — in its more experimental workflows — open the door to automated remediation suggestions. Multiple AI backends are supported beyond OpenAI, including Azure OpenAI, Amazon Bedrock and local models, and sensitive fields can be anonymized before anything leaves the cluster, which helps platform teams adopt it under real-world security policies.
K8sGPT is free and open source under the Apache-2.0 license. If you use a hosted LLM backend for explanations, you pay that provider’s normal API costs; local backends are free.
K8sGPT is for developers and SRE teams who run Kubernetes and want faster, friendlier troubleshooting — especially teams where not everyone is a Kubernetes veteran.