Using connector templates

Connector templates are predefined Integration Service connector configurations for common LLM providers. When you create a custom connector through AI Trust Layer, selecting a template pre-populates the connector with the required API resources, authentication setup, and JavaScript hooks for that provider, reducing manual configuration.

Create a custom connector from a template

1. In Automation Suite, navigate to AI Trust Layer > LLM configurations.
2. Wählen Sie Konfiguration hinzufügen aus.
3. Select the Tenant, Product, and Feature values for the configuration.
4. Under Model Configuration, configure the LLM Name and API Type fields.
5. In the Connector field, select Create custom connector.
6. In the Create custom connector dialog, select one of the available templates.
7. Select Create connector. You are redirected to Connector Builder, where you can edit the configuration, save it, and publish the connector.

Ergebnis

The custom connector is created and pre-configured based on the selected template. You can edit the connector's JavaScript hooks to further customize the LLM request and response behavior.

Available templates

The templates available in the Create custom connector dialog depend on the API Type you selected in the model configuration.

AWS Bedrock

Available for API type: AWS Bedrock

When to use: Use this template when connecting to your own Amazon Bedrock LLM instance and you need to control or modify LLM request and response attributes.

Authentication: This template follows the standard AWS Bedrock authentication, based on API key.

Azure OpenAI

Available for API type: OpenAI

When to use: Use this template when connecting to your own Azure OpenAI LLM instance and you need to control or modify LLM request and response attributes.

Authentication: This template follows the standard Azure OpenAI authentication.

Google Vertex AI

When to use: Use this template when connecting to your own Google Vertex AI LLM instance and you need to control or modify LLM request and response attributes.

Authentication: This template follows the standard Google Vertex AI authentication, based on API key.

OpenAI V1 Compatible

Available for API type: OpenAI

When to use: Use this template when connecting to your own OpenAI or OpenAI V1 Compatible instance — for example, Fireworks — and you need to control or modify LLM request and response attributes.

Authentication: This template follows the standard OpenAI V1 Compatible authentication, based on API key.

Custom AI Gateway

Available for API type: All

When to use: Use this template when your LLMs are behind an AI Gateway.

Authentication: For configuration details, refer to the Connector Builder authentication documentation.

The template includes two API resources required for the Custom authentication type:

• /onProvision — Called when the Integration Service connection is created. Fetches a temporary token from your Identity Provider and saves it on the connection together with an expiration time.
• /onRefresh — Used to refresh the token. The target URL, pre-request hook, and post-request hook must match those configured for /onProvision.

If you choose an authentication type other than Custom, remove the /onRefresh and /onProvision API resources from the connector.

Edit the JavaScript hooks

After creating a connector from a template, you can customize its behavior by editing the pre-request and post-request JavaScript scripts. For details, refer to the Connector Builder documentation.

Before editing the scripts, review the available code snippets and the documentation for each reserved keyword.