DevOps AI Toolkit

# kagent Setup Guide **Connect DevOps AI Toolkit MCP Server to kagent agents.** ## Prerequisites - [kagent](https://kagent.dev) installed in your Kubernetes cluster - DevOps AI Toolkit MCP server deployed (see [Kubernetes Setup](kubernetes-setup.md)) ## Setup Steps ### Step 1: Create RemoteMCPServer Resource Create a `RemoteMCPServer` resource pointing to the DevOps AI Toolkit MCP server: ```yaml apiVersion: kagent.dev/v1alpha2 kind: RemoteMCPServer metadata: name: dot-ai-mcp namespace: kagent spec: description: DevOps AI Toolkit MCP Server protocol: STREAMABLE_HTTP sseReadTimeout: 5m0s terminateOnClose: true timeout: 30s url: http://dot-ai-mcp.dot-ai.svc.cluster.local:3456 ``` **Note**: Adjust the URL if you deployed to a different namespace or service name. ### Step 2: Create or Update Agent Create a new agent that uses the DevOps AI Toolkit MCP server: ```yaml apiVersion: kagent.dev/v1alpha2 kind: Agent metadata: name: devops-agent namespace: kagent spec: type: Declarative description: DevOps automation agent with AI-powered Kubernetes tools declarative: modelConfig: default-model-config # Use your existing model config stream: true systemMessage: You are a DevOps automation agent with access to AI-powered Kubernetes deployment, troubleshooting, and project setup tools. tools: - type: McpServer mcpServer: apiGroup: kagent.dev kind: RemoteMCPServer name: dot-ai-mcp toolNames: - recommend - version - manageOrgData - remediate - projectSetup ``` **Or update an existing agent** to include DevOps AI Toolkit tools: ```yaml kubectl patch agent my-agent -n kagent --type=merge -p ' spec: declarative: tools: - type: McpServer mcpServer: apiGroup: kagent.dev kind: RemoteMCPServer name: dot-ai-mcp toolNames: - recommend - version - manageOrgData - remediate - projectSetup ' ``` ### Step 3: Verify Setup Check that the RemoteMCPServer discovered all tools: ```bash kubectl get remotemcpserver dot-ai-mcp -n kagent -o yaml ``` You should see 5 tools in `status.discoveredTools`: `recommend`, `version`, `manageOrgData`, `remediate`, `projectSetup`. Check that the agent is ready: ```bash kubectl get agent devops-agent -n kagent ``` Status should show `READY: True`. ## Troubleshooting **Agent pod not starting:** - Check if `toolNames` is specified (kagent requires explicit tool list) - Verify the MCP server URL is accessible from the kagent namespace - Check agent pod logs: `kubectl logs -n kagent <agent-pod>` **Tools not discovered:** - Verify RemoteMCPServer status: `kubectl describe remotemcpserver dot-ai-mcp -n kagent` - Check MCP server is running: `kubectl get pods -n dot-ai` - Test MCP server directly: `curl http://dot-ai-mcp.dot-ai.svc.cluster.local:3456` **Tool invocation fails:** - Ensure capability scan completed successfully - Check MCP server logs: `kubectl logs -n dot-ai deployment/dot-ai-mcp` - Verify network connectivity between kagent and dot-ai namespaces