flux-operator

The Flux Operator MCP Server provides an AI-enabled interface for interacting with the Flux Operator running in your Kubernetes clusters. It leverages Flux and the operator's capabilities to manage declarative GitOps workflows, generate real-time insights about deployment status, and surface reconciliation metrics. You can use the server to reason about cluster state, view Flux controller health, and issue prompts that translate into Flux actions or queries against Flux resources. The server exposes conversational prompts that help analyze deployment across environments, troubleshoot issues, and perform operations related to Flux configurations and multi-tenant setups.

Typical use cases include requesting current reconciler status, inspecting Flux CRDs versions in a cluster, examining the readiness of Flux components, and guiding the creation or debugging of ResourceSet-based templates for self-service environments. The server acts as a bridge between AI-assisted workflows and GitOps pipelines, enabling more natural interactions with the Flux-based delivery pipelines directly from your MCP tooling.

Prerequisites:

• A Kubernetes cluster with kubectl configured
• Helm installed (optional but recommended for Flux Operator installation)
• Access to install CRDs and manage Flux resources

1. Install Flux Operator and Flux controllers in your cluster (example using Helm):

helm repo add fluxoperator https://fluxoperator.dev/helm
helm install flux-operator fluxoperator/flux-operator --namespace flux-system --create-namespace

2. Verify Flux Operator installation:

kubectl get pods -n flux-system

3. Run the MCP server for Flux Operator (as described in this repository's mcp_config):

docker run -i ghcr.io/controlplaneio-fluxcd/flux-operator:latest

4. Configure access to your cluster from the MCP server as needed (e.g., kubeconfig or in-cluster configuration). Ensure the MCP server has network access to the Kubernetes API server and the appropriate RBAC permissions to read Flux resources and CRDs.

5. Optional: customize environment and watch scope if you want to limit the MCP server to a specific namespace or set additional environment variables.

Notes and tips:

• The Flux Operator provides extensive CRDs (FluxInstance, FluxReport, etc.). Ensure CRDs are installed in your cluster prior to using the MCP server.
• If you run the MCP server outside the cluster, ensure the server has proper kubeconfig context or API access to the target cluster.
• Environment variables such as WATCH_NAMESPACE can be used to scope operations to a single namespace; unset to monitor all namespaces.
• When troubleshooting, verify Flux controller readiness and ensure the Flux components are up-to-date with the desired distribution version.
• The MCP server leverages Flux metrics and the Flux Web UI for observability; consider exposing these surfaces securely (Ingress/SSO) for production use.