Tenant Management

Tenants are the primary isolation boundary in MCTL. Each tenant gets its own Kubernetes namespace, RBAC policies, and resource quotas.

Create a Tenant

Via MCP (Natural Language)

"Create a new tenant called backend-team"

This calls the mctl_create_tenant tool, which submits a workflow that:

1. Creates a namespace in the cluster
2. Sets up RBAC roles and bindings
3. Configures resource quotas and limit ranges
4. Registers the tenant in the service catalog

Via Self-Service Form

Visit mctl.ai and submit a workspace request after GitHub sign-in. Confirm provisioning and MCTL sign-in with your operator before using it; the request session does not itself grant platform UI access.

List Tenants

Via MCP

"Who am I on MCTL?"
"Show me details of my-team"

mctl_whoami reports your accessible namespaces; mctl_get_tenant reads a specific tenant. Listing all tenants with mctl_list_tenants requires platform admin access.

Via API (platform admins only)

curl -H "Authorization: Bearer $TOKEN" https://api.mctl.ai/api/v1/tenants

Get Tenant Details

"Show me details of the staging tenant"
"What services are running in production?"

This returns tenant configuration and services. Use the resource, domain and operations tools separately for those views.

Delete a Tenant

DANGER

Deleting a tenant removes the namespace and all its resources. This action cannot be undone.

"Delete the test-env tenant"

The mctl_delete_tenant tool requires the tenant owner role (or platform admin) and submits a workflow to:

1. Remove all services in the tenant
2. Delete the namespace and all resources
3. Clean up RBAC bindings and catalog entries

Tenant Naming

Tenant namespaces use the team name directly (no prefix). For example, a tenant named backend-team creates a namespace called backend-team.

Resource Quotas

Defaults depend on the provisioning entry point. The operations API currently supplies the following defaults; existing tenant values and portal requests may override them. Inspect the actual workspace quota before sizing a workload:

Resource	Default
CPU requests	500m
CPU limits	3 cores
Memory requests	1280Mi
Memory limits	4Gi
Pods	10