# serve The serve commands allow you to run k8sgpt in a grpc server mode. This would be enabled typically through `k8sgpt serve` and is how the in-cluster k8sgpt deployment functions when managed by the [k8sgpt-operator](https://github.com/k8sgpt-ai/k8sgpt-operator) The grpc interface that is served is hosted on [buf](https://buf.build/k8sgpt-ai/schemas) and the repository for this is [here](https://github.com/k8sgpt-ai/schemas) ## grpcurl A fantastic tool for local debugging and development is `grpcurl` It allows you to form curl like requests that are http2 e.g. ``` grpcurl -plaintext -d '{"namespace": "k8sgpt", "explain" : "true"}' localhost:8080 schema.v1.ServerService/Analyze ``` ``` grpcurl -plaintext localhost:8080 schema.v1.ServerService/ListIntegrations { "integrations": [ "trivy" ] } ``` ``` grpcurl -plaintext -d '{"integrations":{"trivy":{"enabled":"true","namespace":"default","skipInstall":"false"}}}' localhost:8080 schema.v1.ServerService/AddConfig ```