Skip to main content

Google Cloud Vertex AI

App Vendor: Google Cloud Vertex AI

App Category: AI Models

Connector Version: 1.0.1

API Version: v1

About App

The Google Cloud Vertex AI app helps security teams integrate generative AI into their workflows. It enables automation, data enrichment, and advanced analysis to support faster, more informed security decisions.

The Google Cloud Vertex AI app is configured with Cyware Orchestrate to perform the following actions:

Action Name

Description

Generate Content

This action creates content by invoking a Vertex AI generative model with multimodal inputs.

Generic Action

This is a generic action used to make requests to any Google Cloud Vertex AI endpoint.

Configuration Parameters

The following configuration parameters are required for the Google Cloud Vertex AI app to communicate with the Google Cloud Vertex AI enterprise application. The parameters can be configured by creating instances in the app.

Parameter

Description

Field Type

Required/Optional

Comments

Project ID 

Enter the Google Cloud project ID to identify the owner of the Vertex AI resources.

Text

Required

Service Account Key 

Enter the contents of the Google Cloud Service Account JSON key file required to authenticate requests to Vertex AI. 

Password

Required

You must create the service account in Google Cloud to obtain this key. For more information, see Create a service account.

Verify 

Choose your preference to verify SSL or TLS while making requests. It is recommended to set this option to yes. Passing no may result in connection errors.

Boolean

Optional

By default, verification is enabled.

Timeout 

Enter the timeout value in seconds. This is the number of seconds that requests will wait to establish a connection with Google Cloud Vertex AI.

Integer

Optional

Allowed range:

15-120

Default value:

15

Temperature 

Enter the sampling temperature to control the randomness of the output. Higher values make the output more random, while lower values make it more focused and deterministic.

Float

Optional

Allowed range:

0 to 2

Default value:

1

Max Output Tokens 

Enter the maximum number of tokens the model can generate in the response before stopping.

Integer

Optional

The maximum value varies depending on the model.

Top P (Nucleus Sampling) 

Enter the top p-value to control how focused or varied the generated text is. Lower values produce more focused output, while higher values allow more variety.

Float

Optional

Allowed range:

0.0 to 1.0

Action: Generate Content

This action creates content by invoking a Vertex AI generative model with multimodal inputs.

Action Input Parameters

Parameter

Description

Field Type

Required/Optional

Comments

Prompt

Enter the prompt for the model to generate a response.

Example:

Provide recommended response actions for a potential intrusion attempt.

Text

Required

Published Model

Enter the fully qualified name of the publisher model to generate content.

Example:

publishers/google/models/gemini-2.5-flash-lite.

Text

Optional

Allowed format: publishers/*/models/*

This field is required if the endpoint ID is not provided. Note that only the published models that support generative capabilities can be used here.

Endpoint ID

Enter the specific endpoint ID if you are invoking a model deployed to an endpoint.

Text

Optional

This is required if the published model is not provided. Note that only the endpoints that support generative models can be used here.

Location

Enter the region name to specify where the Vertex AI model is deployed.

Example:

us-central1

Text

Optional

By default, it is set to global.

Cached Content

Enter the full resource name of cached content to use as additional context for the request.

Example:

projects/{project}/locations/{location}/cachedcontents/{cached_content_id}

Text

Optional

Labels

Enter the labels as key-value pairs to associate metadata for billing and reporting purposes.

Key Value

Optional

Safety Settings

Enter the safety settings to define harm categories and blocking thresholds for sanitizing model responses.

Key Value

Optional

Response MIME type

Enter the MIME type to define the format of the model's raw response.

Text

Optional

Allowed types:

text/plain for text output and application/json for JSON responses.

By default, it is set to application/json.

Example Request

[
  {
    "labels": {
      "title": "billing tracker for generate Content"
    },
    "prompt": "Provide a brief rundown of cyware products",
    "published_model": "publishers/google/models/gemini-2.5-flash-lite-preview-09-2025",
    "safety_settings": {},
    "response_mime_type": "text/plain"
  }
]

Action: Generic Action

This is a generic action used to make requests to any Google Cloud Vertex AI endpoint.

Action Input Parameters 

Parameter

Description

Field Type

Required/Optional

Comments

Full URL 

Enter the full URL along with the path of the Vertex AI endpoint to make the request.

Example:

https://aiplatform.googleapis.com/v1/projects/{project}/locations/{location}/cachedcontents

Text

Required

Method 

Enter the HTTP method to make the request.

Text

Required

Allowed values:

GET, PUT, POST, and DELETE

Query Params 

Enter the query parameters to pass to the API.

Key Value

Optional

Payload 

Enter the payload to pass to the API.

Any

Optional

Extra Fields 

Enter the extra fields to pass to the API.

Key Value

Optional

Allowed keys:

payload_json, download, files, filename, retry_wait, retry_count, custom_output, and response_type

Example Request 

[
  {
    "method": "GET",
    "base_url": "https://aiplatform.googleapis.com",
    "endpoint": "$discovery/rest",
    "extra_fields": {},
    "query_params": {
      "version": "v1"
    }
  }
]