ks_create_cluster

Instructions

Implementation Reference

• src/tools/kubernetes/index.ts:24-31 (handler)

  The handler function for the ks_create_cluster tool. It calls assertWriteAllowed to check write permissions, then sends a POST request to the IBM Cloud Kubernetes API endpoint /v2/createCluster with the provided parameters (name, datacenter, machine_type, worker_num, kube_version, provider, vpc_id, subnet_id, resource_group_id). Uses safeTool wrapper for error handling.

  }, async (p) => safeTool(async () => { w();
    return client.post(`${base}/v2/createCluster`, {
      name:p.name, dataCenter:p.datacenter, machineType:p.machine_type,
      workerNum:p.worker_num||1, kubeVersion:p.kube_version,
      provider:p.provider||"vpc-gen2", vpcID:p.vpc_id, subnetID:p.subnet_id,
      resourceGroup:p.resource_group_id,
    });
  }));

• src/tools/kubernetes/index.ts:18-23 (schema)

  The schema/input parameter definitions for ks_create_cluster. Defines the Zod validation schema with required fields (name, datacenter, machine_type) and optional fields (worker_num, kube_version, resource_group_id, provider with enum ['classic','vpc-gen2'], vpc_id, subnet_id).

  server.tool("ks_create_cluster", "Create a new Kubernetes cluster", {
    name: z.string(), datacenter: z.string().describe("Zone e.g. dal10"),
    machine_type: z.string().describe("Worker machine type"), worker_num: z.number().optional(),
    kube_version: z.string().optional(), resource_group_id: z.string().optional(),
    provider: z.enum(["classic","vpc-gen2"]).optional(),
    vpc_id: z.string().optional(), subnet_id: z.string().optional(),

• src/tools/kubernetes/index.ts:18-31 (registration)

  The tool registration via server.tool() call. The ks_create_cluster tool is registered inside the registerKubernetesTools function (line 7) which is called from src/server.ts line 56. The tool name is 'ks_create_cluster' with description 'Create a new Kubernetes cluster'.

  server.tool("ks_create_cluster", "Create a new Kubernetes cluster", {
    name: z.string(), datacenter: z.string().describe("Zone e.g. dal10"),
    machine_type: z.string().describe("Worker machine type"), worker_num: z.number().optional(),
    kube_version: z.string().optional(), resource_group_id: z.string().optional(),
    provider: z.enum(["classic","vpc-gen2"]).optional(),
    vpc_id: z.string().optional(), subnet_id: z.string().optional(),
  }, async (p) => safeTool(async () => { w();
    return client.post(`${base}/v2/createCluster`, {
      name:p.name, dataCenter:p.datacenter, machineType:p.machine_type,
      workerNum:p.worker_num||1, kubeVersion:p.kube_version,
      provider:p.provider||"vpc-gen2", vpcID:p.vpc_id, subnetID:p.subnet_id,
      resourceGroup:p.resource_group_id,
    });
  }));

• src/lib/utils.ts:70-77 (helper)

  The safeTool helper function wraps the handler to catch errors and return proper MCP text content responses (success or error). Also, assertWriteAllowed (line 14-18) is called as w() inside the handler to enforce read-only mode when writes are disabled.

  export async function safeTool<T>(fn: () => Promise<T>): Promise<ReturnType<typeof successContent> | ReturnType<typeof errorContent>> {
    try {
      const result = await fn();
      return successContent(result);
    } catch (error) {
      return errorContent(error);
    }
  }

• src/config.ts:26-37 (helper)

  The KUBERNETES endpoint constant used by the tool. The base URL is 'https://containers.cloud.ibm.com/global', and the tool appends '/v2/createCluster' to it.

  export const IBM_ENDPOINTS = {
    IAM: "https://iam.cloud.ibm.com",
    IAM_IDENTITY: "https://iam.cloud.ibm.com/v1",
    IAM_ACCESS_GROUPS: "https://iam.cloud.ibm.com/v2",
    IAM_POLICY: "https://iam.cloud.ibm.com/v1",
    USER_MANAGEMENT: (accountId: string) =>
      `https://user-management.cloud.ibm.com/v2/accounts/${accountId}`,
  
    VPC: (region: string) =>
      `https://${region}.iaas.cloud.ibm.com/v1`,
  
    KUBERNETES: "https://containers.cloud.ibm.com/global",