Management resource workflows
Work with tenants, services, instances, backups, and region records without confusing resource IDs or lifecycle states.
The management service exposes resources used to operate deployed Minds. It is a separate application from the signed-in dashboard. Set MINDS_MANAGEMENT_URL to the origin supplied for your deployment and use its configured authentication.
Tenants
A tenant is the management record that owns an instance's endpoint, plan, region, configuration, and lifecycle. An organization ID scopes access; it is not interchangeable with the tenant ID.
| Task | Method and path | Result |
|---|---|---|
| List accessible records | GET /api/v1/tenants | Validated list response; query fields include organization, filters, and pagination defined by ListTenantsQuery |
| Create | POST /api/v1/tenants | 202 accepted with CreateTenantResponse; provisioning can still be in progress |
| Read | GET /api/v1/tenants/[id] | Tenant record |
| Change configuration | PATCH /api/v1/tenants/[id] | Updated record following schema validation |
| Inspect state | GET /api/v1/tenants/[id]/status | Status and health response |
| Request scaling | POST /api/v1/tenants/[id]/scale | Deployment-specific scaling operation |
| Remove | DELETE /api/v1/tenants/[id] | Deletion operation; inspect its endpoint response and retention behavior before use |
A management create request requires org_id, name, and plan. The schema also accepts optional project_id, slug, region, config, service_type, bundle, and service_config.
{
"org_id": "<accessible-organization-id>",
"name": "Research memory",
"plan": "free",
"region": "<region-returned-by-your-catalog>"
}Use the tenant endpoint contract for all required fields, default values, and result types. Older schema enum values such as starter and legacy storage fields remain visible in the reference for compatibility; they are not current packaging recommendations.
Services and instances
A service record represents a provisioned service type. An instance is the running deployment to which services may be bound. Keep their IDs separate.
/api/v1/services: list or create service records./api/v1/services/[id]: read, update with PUT, or delete a service./api/v1/services/[id]/bind: bind a service to an instance./api/v1/services/[id]/status: read service status./api/v1/instances: create an instance or list by requiredtenant_idquery./api/v1/instances/[instanceId]: inspect or delete an instance./api/v1/instances/[instanceId]/services/{enable,disable,status}: control or inspect enabled services./api/v1/tenants/[id]/services/{enable,disable}: tenant-scoped service controls./api/v1/bundles/expand: resolve a bundle definition into its configured service selection.
The instance-create route uses tenantId, instanceType, and tier in its JSON body. This differs from the snake-case tenant-create contract. The list-instances route returns a raw array; do not assume every list has a {data:...} wrapper.
Backups and restore
List existing backups before making a recovery decision:
curl --fail-with-body \
--header "x-api-key: $MINDS_MANAGEMENT_API_KEY" \
"$MINDS_MANAGEMENT_URL/api/v1/tenants/$MINDS_TENANT_ID/backups"POST to that collection requests a backup and returns 202. Restore uses POST /api/v1/tenants/[id]/backups/[backupId]/restore; deleting a backup uses DELETE on its resource path. A queued backup or restore is not a completed recovery. Inspect the returned operation and later state before depending on it.
The complete backup request and response schemas appear in Schema reference. These management backups are different from local engine backup commands and HMR recovery-material confirmation.
Region catalog and health
Read regions through /api/v1/regions; use /api/v1/regions/select to apply the deployment's selection logic. Region resources also expose read/update/delete and health-check routes. Region mutations are operational administration, not a customer provisioning shortcut.
Current region-health source derives several booleans from stored region status. The dashboard may additionally return a catalog fallback. Do not treat these flags as proof of a live Kubernetes, ingress, or certificate probe. See proxy behavior.
Complete contracts
The endpoint catalog includes all management routes, auth resolution, request schemas, status branches, and helper dependencies. Read Errors and responses for retry and asynchronous-operation handling.