Save All Integrations

Endpoint: POST /api/integrations

Creates or updates one or more integrations in bulk. Each integration in the array is upserted along with its associated models or APIs. If an integration already exists, its type cannot be changed.

Password-type configuration values (such as api_key and password) are automatically stored as secrets.

Request body

Format the request body as an array of integration objects:

Parameter	Description	Type	Required/ Optional
name	The unique name for the integration provider. Cannot be changed after creation.	string	Required.
type	The integration type. Cannot be changed after creation. Supported values: ollama azure_openai openai perplexity grok cohere mistral anthropic vertex_ai vertex_ai_gemini huggingface aws_bedrock_anthropic aws_bedrock_cohere aws_bedrock_titan pineconedb weaviatedb pgvectordb mongovectordb amqp kafka nats aws_sqs azure_service_bus gcp_pubsub ibm_mq aws gcp relational_db sendgrid git	string	Required.
category	The category of the integration. Supported values: AI_MODEL: For AI/LLM integrations. VECTOR_DB: For Vector database integrations. MESSAGE_BROKER: For message broker integrations. CLOUD: For cloud provider integrations. RELATIONAL_DB: For relational database integrations. GIT: For Git repository integrations. EMAIL: For email provider integrations.	string	Optional.
description	A description for the integration.	string	Optional.
enabled	Whether the integration is active and available for use.	boolean	Optional.
configuration	A key-value map of provider-specific configuration, such as API keys and endpoints. Keys and required values vary by integration type. See Configuration keys by integration type.	object	Optional.
tags	A list of key-value tag objects for organizing or filtering integrations. Each tag contains a key (string) and value (string).	array	Optional.
apis	A list of models or APIs to associate with this integration provider. See Integration model object.	array	Optional.

Integration model object

While adding integrations for AI/LLMs, Vector databases or Relational databases, you must add the associated models, indexes, or tables, respectively. They must be formatted within the apis field as an array of the following object:

Parameter	Description	Type	Required/ Optional
apis.integrationName	Name of the parent integration provider to which the model must be attached to. Must match the enclosing integration's name.	string	Required.
apis.api	Name of the specific model or API under the provider (e.g., `gpt-4o`, `text-embedding-3-small`).	string	Required.
apis.description	A human-readable description of this model or API.	string	Required.
apis.enabled	Whether this model/API is active and available for use.	boolean	Optional.

For example, if you are adding an OpenAI integration with gpt-4o as the model; format the request body as follows:

{
      "name": "<YOUR-INTEGRATION-NAME>",
      "type": "openai",
      "category": "AI_MODEL",
      "description": "<YOUR-DESCRIPTION>",
      "enabled": true,
      "configuration": {
        "api_key": "<YOUR-API-KEY>"
      },
      "apis": [
        {
          "integrationName": "<YOUR-INTEGRATION-NAME>",
          "api": "gpt-4o", // model name
          "description": "GPT-4o model", // model description
          "enabled": true
        }
      ]
    }

Configuration keys by integration type

The configuration object is a key-value map of provider-specific configuration, such as API keys and endpoints. Keys and required values vary by integration type.

Parameter	Required keys	Optional keys
ollama	endpoint	header, api_key
azure_openai	api_key, endpoint	-
openai	api_key	endpoint,organizationId
perplexity	api_key	–
grok	api_key	-
cohere	api_key, endpoint	–
mistral	api_key, endpoint	–
anthropic	api_key, endpoint	completionsPath, version, betaVersion
vertex_ai	projectName, environment, publisher, file	–
vertex_ai_gemini	projectName, environment, file	–
huggingface	api_key, namespace	–
aws_bedrock_anthropic	connectionType, region Based on the `connectionType` value, the following additional keys may be required: ACCESS_KEY: Requires user (access key) and api_key (access secret). EXTERNAL_ROLE: Requires roleArn and externalId. CONDUCTOR_ROLE: No additional keys required.	awsAccountId
aws_bedrock_cohere	connectionType, region Based on the `connectionType` value, the following additional keys may be required: ACCESS_KEY: Requires user (access key) and api_key (access secret). EXTERNAL_ROLE: Requires roleArn and externalId. CONDUCTOR_ROLE: No additional keys required.	awsAccountId
aws_bedrock_titan	connectionType, region Based on the `connectionType` value, the following additional keys may be required: ACCESS_KEY: Requires user (access key) and api_key (access secret). EXTERNAL_ROLE: Requires roleArn and externalId. CONDUCTOR_ROLE: No additional keys required.	awsAccountId
pineconedb	api_key, projectName, environment	–
weaviatedb	api_key, endpoint	–
pgvectordb	user, password, datasourceURL, dimensions, distance_metric, indexing_method	inverted_list_count
mongovectordb	endpoint, namespace, dimensions, distance_metric	inverted_list_count
amqp	protocol, user, password, endpoint, port, namespace	–
kafka	endpoint, connectionType, protocol	groupId
nats	endpoint, connectionType, authenticationType, protocol, tls	–
aws_sqs	connectionType, region Based on the `connectionType` value, the following additional keys may be required: ACCESS_KEY: Requires user (access key) and api_key (access secret). EXTERNAL_ROLE: Requires roleArn and externalId. CONDUCTOR_ROLE: No additional keys required.	awsAccountId
azure_service_bus	connectionType	endpoint, namespace
gcp_pubsub	projectName, location, file	–
ibm_mq	endpoint, port, queueManager, channel, protocol, pubSubMethod, authenticationType, tls	–
aws	connectionType, region Based on the `connectionType` value, the following additional keys may be required: ACCESS_KEY: Requires user (access key) and api_key (access secret). EXTERNAL_ROLE: Requires roleArn and externalId. CONDUCTOR_ROLE: No additional keys required.	awsAccountId
gcp	projectName, environment, file	–
relational_db	jdbcDriver, user, password, datasourceURL	–
sendgrid	api_key	–
git	user, api_key	–

For example, if you are adding an Anthropic Claude integration, the configuration looks like this:

{
      "name": "<YOUR-INTEGRATION-NAME>",
      "type": "anthropic",
      "category": "AI_MODEL",
      "description": "<YOUR-DESCRIPTION>",
      "enabled": true,
      "configuration": {
        "api_key": "<YOUR-API-KEY>",
        "endpoint": "https://api.anthropic.com/v1"
      },
      "apis": [
        {
          "integrationName": "<YOUR-INTEGRATION-NAME>",
          "api": "claude-sonnet-4-6", // model name
          "description": "claude-sonnet-4-6 model", // model description
          "enabled": true
        }
      ]
    }

Response

Status	Description
200 OK	Indicates that the integration is saved successfully.
400 Bad Request	Returns for an invalid input. For example, attempting to change the `type` of an existing integration.
401 Unauthorized	Indicates that the request is not authenticated.
403 Forbidden	Indicates that the authenticated user does not have permission to update one or more integrations.

Examples

Save all integrations

Request

curl -X 'POST' \
  'https://<YOUR-SERVER-URL>/api/integrations/' \
  -H 'accept: */*' \
  -H 'X-Authorization: <TOKEN>' \
  -H 'Content-Type: application/json' \
  -d '[
    {
      "name": "open-ai-test",
      "type": "openai",
      "category": "AI_MODEL",
      "description": "OpenAI integration for LLM tasks",
      "enabled": true,
      "configuration": {
        "api_key": "sk-..."
      },
      "apis": [
        {
          "integrationName": "open-ai-test",
          "api": "gpt-4o",
          "description": "GPT-4o model",
          "enabled": true
        }
      ]
    },
    {
      "name": "sendgrid-test",
      "type": "sendgrid",
      "category": "EMAIL",
      "description": "SendGrid integration for email tasks",
      "enabled": true,
      "configuration": {
        "api_key": "SG...."
      },
"apis": []
    }
  ]'

Response

Returns 200 OK, indicating that the integrations have been saved successfully.