OpenScale kinds

The openscale service has 8 resource kinds, each listed below with its endpoints, fields, and dependencies. Values resolve late, so you reference another resource with ${kind.ref_name}. For defaults, validation rules, and examples, run wxctl explain <kind>.

YAML envelope & reference syntax

• Envelope: A config is one or more YAML documents separated by ---. Each has top-level kind and ref_name, then the fields below at the top level (not nested under spec).
• ref_name: Unique handle for this resource within the config. Used to reference it from other resources, then stripped before the API call (not a schema field).
• References: Reference another resource by its ref_name: ${<kind>.<ref_name>} resolves to its id, ${<kind>.<ref_name>.<field>} to a specific field. Values resolve late, at plan/apply time.

​

data_mart

An OpenScale data mart is the per-instance datastore that holds monitoring records (payload, feedback, metrics) for every subscribed deployment. Runs on SaaS + Software · ID field id Endpoints Fields Computed outputs: entity, metadata. Consumers

• data_set → .data_mart_id (required)
• monitor_instance → .data_mart_id (required)
• subscription → .data_mart_id (required)

​

data_set

A data set is a typed store within an OpenScale data mart (payload logging, feedback, or business payload) that records the data used to compute a subscription’s monitor metrics. Runs on SaaS + Software · ID field id Endpoints Fields Computed outputs: entity, metadata. Depends on

• data_mart_id → data_mart (required)
• target.target_id → subscription (required)

​

guardrails_policy

A guardrails policy defines content-filtering and safety rules (such as PII or HAP detection) that OpenScale enforces on generative-AI model inputs and outputs. Runs on SaaS + Software · ID field id Endpoints Fields Computed outputs: entity, metadata.

​

integrated_system

An integrated system registers an external service (notifications, a custom metrics provider, or a third-party endpoint) that OpenScale calls during monitoring. Runs on SaaS + Software · ID field id Endpoints Fields Computed outputs: entity, metadata.

​

monitor_definition

A monitor definition registers a custom monitor type in OpenScale, declaring its metrics, thresholds, and tags. Runs on SaaS + Software · ID field id Endpoints Fields Computed outputs: entity, metadata. Consumers

• monitor_instance → .monitor_definition_id (optional)

​

monitor_instance

A monitor instance activates a monitor (quality, fairness, drift, explainability, or a custom definition) on a subscription, with its parameters and alert thresholds. Runs on SaaS + Software · ID field id Endpoints Fields Computed outputs: entity, metadata. Depends on

• data_mart_id → data_mart (required)
• target.target_id → subscription (required)
• monitor_definition_id → monitor_definition (optional)

​

service_provider

A service provider binds a machine-learning serving environment (such as a watsonx.ai or WML space) to OpenScale so its deployments can be discovered and monitored. Runs on SaaS + Software · ID field id Endpoints Fields Computed outputs: entity, metadata. Depends on

• deployment_space_id → space (optional)

Consumers

• subscription → .service_provider_id (required)

​

subscription

A subscription connects one deployed model to OpenScale for monitoring, linking it to a service provider and data mart and capturing its input and output schema. Runs on SaaS + Software · ID field id Endpoints Fields Computed outputs: entity, metadata. Depends on

• asset.asset_id → wml_deployment (required)
• data_mart_id → data_mart (required)
• deployment.deployment_id → wml_deployment (required)
• service_provider_id → service_provider (required)

Consumers

• data_set → .target.target_id (required)
• monitor_instance → .target.target_id (required)