Platform Capabilities¶

This page describes AI Studio’s externally available programmatic capabilities for working with data: the Operator API, SQL connection strings, REST data sources, Webhooks, and more.

The documentation repository does not currently provide a complete OpenAPI catalog for application-level REST APIs such as knowledge bases, agents, and workflows. This page indexes the capabilities and describes their invocation patterns. Refer to product release notes, console API examples, or a future OpenAPI specification for exact endpoints and fields. For procedures, see Developer. For connector fields, see Connectors.

Capability Overview¶

Capability	Direction	Purpose	Authentication and Billing
Operator API	Invoke processing capabilities	Parsing, segmentation, embedding, information extraction, and more	Genesis API Key authentication and Credit billing are available by default; OAuth 2.0 or public access can also be configured
SQL Connection String	Direct external queries against the Catalog	Queries from BI tools, clients, and third-party applications	Catalog connection credentials; queries consume query compute resources
REST Data Sources and Webhooks	Catalog ingestion / export	Load from third-party REST APIs; deliver processing results or events	Connector credentials or recipient-side signature verification

Requirement	See
Invoke model inference	Model API
Invoke an individual data processing task	Operator API
Query the Catalog through a connection string	SQL Connection String
External HTTP integration or result delivery	REST Data Sources and Webhooks; Data Ingestion

The data capabilities above all operate within the scope of the workspace’s Catalog and compute resources.

Operator API¶

Publish a system operator or custom operator as a REST service. Once published, it can be invoked externally and used as an agent tool. Publishing and configuration are completed in the Catalog.

Invocation Conventions¶

Item	Description
Invocation mode	Synchronous: returns a result after receiving the request parameters; the operator layer does not provide a separate asynchronous job model
Long-running tasks	Orchestrate them as workflows, then obtain results through job status or a Webhook
Input parameters	Correspond to the operator’s input and parameter signature (name, type, whether required, and default value); input sources depend on the publishing configuration

Authentication and Runtime¶

Item	Description
Authentication	API Key (default; a Genesis key can be reused), OAuth 2.0, or public access
Billing and rate limits	When using an API Key, billing and rate-limit policies are usually shared with the Genesis key and Credit ledger
Runtime resources	Model operators can invoke the model API; other operators are bound to compute resources
Runtime configuration	Timeout, maximum concurrency, invocation frequency, and other settings can be adjusted in the service details

Request Example¶

POST https://<OPERATOR_ENDPOINT>
Authorization: Bearer <API_KEY>
Content-Type: application/json

{
  "input": "..."
}

Service URLs and input/output parameters vary by operator. Refer to the description under API Service on the operator details page.

SQL Connection String¶

Provides standard database connection information so clients, BI tools, and drivers that support the MySQL protocol can directly query databases and tables in the Catalog.

Elements¶

Element	Description
host / port	Query service address and port
Database / table path	Target database (and table)
Authentication	Username and password, or equivalent credentials

Refer to the actual connection string generated in the Catalog for the exact values.

Connection Examples¶

mysql -h <HOST> -P <PORT> -u <USER> -p

host=<HOST>  port=<PORT>  user=<USER>  password=<PASSWORD>  database=<DATABASE>

Use Cases¶

Suitable	Not Suitable
Direct report connections, interactive exploration, and on-demand retrieval	High-frequency, highly concurrent programmatic data consumption (prefer materializing workflow results in a table or using the Operator API)

SQL queries are usually billed by the online duration of query compute resources, rather than priced separately as a per-call SQL API (refer to the billing information). For syntax, see SQL Reference.

REST Data Sources and Webhooks¶

REST Data Sources¶

Use a REST API or GraphQL connector to integrate with a third-party HTTP endpoint and write data into the Catalog through a loading task. For connector fields, see Connectors.

Configuration	Description
Base URL / GraphQL endpoint	Service root URL
Authentication method	API Key, Bearer Token, OAuth 2.0, or no authentication
Timeout, pagination, data path, and default request headers	Retrieval strategy

Configure the exact path, method, and query parameters in the loading task.

Webhooks¶

Scenario	Description
Data export	Push processing results from the Catalog to an external recipient over HTTP
Platform notifications	Event callbacks for jobs, alerts, and more (subject to product availability)

The recipient should verify the request source, such as through signatures or an IP allowlist. Refer to the actual API for the signing algorithm and payload fields. For recipient implementation, see the relevant integration guidance under Developer.

Data Direction Comparison¶

Capability	Direction
REST data source	External → MOI
Webhook export	MOI → External
SQL connection string	Direct external queries against the Catalog
Operator API	Invoke platform processing capabilities

APIs Not Yet Included¶

The following capabilities may exist in the console or beta environments, but this reference does not yet document each endpoint because no stable OpenAPI specification is available:

Knowledge base management APIs
Agent, skill, and tool management APIs
Workflow triggering and job query APIs
External application hosting APIs

They will be added after official specifications or release notes become available. Before integrating, refer to the API documentation available in the console and to support channels.