(For Agent Only) Gemini Agent Studio Connection Method

1. After specifying Target Details, set Target Type to Agent and Connection Method to Gemini Agent Studio.

2. Configure Endpoint Accessibility. This field indicates if your endpoint is Public or Private (secured within a private network). For Gemini Agent Studio, select Public.
3. Configure the Gemini Agent Studio connection parameters.

   You can find the Gemini Agent Studio connection parameters, GCP project ID, region/location, and agent engine ID in the Gemini Agent Studio and AI Red Teaming Integration setup.

   The agent Resource name contains all required identifiers in the format:

   projects/<PROJECT_NUMBER>/locations/<REGION>/reasoningEngines/<AGENT_ENGINE_ID>

   For Example:

   • Region / Location—the value following locations/ (for example, us-central1).
   • Agent engine ID—the numeric value following reasoningEngines/ (for example, 1112223334445556667).
   • Project number—the numeric value following the leading projects/.

   1. For GCP Project ID, enter your Google Cloud project ID.
   2. For Region/Location, enter the region where your agent is deployed. The value after locations/ in the resource name (for example, us-central1).
   3. For Agent Engine ID, enter the numeric Reasoning Engine ID. The value after reasoningEngines/ in the resource name.
   4. For Service Account Email, enter the email address of the service account you created during GCP IAM setup (format: account-name@YOUR-PROJECT.iam.gserviceaccount.com).
4. Select Next: Advanced Configurations.

   In the Advanced Configurations page, configure Rate Limits and Guardrails/Content Filters.
5. (Optional) Enable Rate Limits for the target endpoint.

   Rate limiting is recommended as your GCP project's Reasoning Engine quota is shared across multiple agents in the same region. Exceeding the per-region quota returns 429 errors that can interrupt a scan.

   1. Specify the Endpoint Rate Limit. This value represents the maximum number of requests per minute allowed for the endpoint.
   2. Specify the Endpoint Rate Limit Error Code.
   3. Provide a Sample Exception JSON.
6. (Optional) Enable Guardrails/Content Filters for output guardrails or content filters on the target endpoint.

   1. Specify the Error Code for Guardrails or Content Filters.
   2. Provide a Sample Exception JSON.
   3. Select Validate Target.

      If validation fails, verify that your GCP IAM is configured correctly: the connector service account has the Service Account Token Creator role scoped to your service account, the custom role includes all six required aiplatform.* permissions, and the agent engine ID and region match the values in your Reasoning Engine resource name.
7. (Mandatory) Configure Target Background.

   AI Red Teaming uses the target background to tailor attacks to the specific use case and risk profile of the AI agent under test.

   Target background information is mandatory for all target types.

   1. Add Industry information.
   2. Add Use Case, describing the specific role of the target, such as customer service or document processing.
   3. (Optional) Select Add Competitor to add the list of Competitors.
   4. Enable Agentic Profiling.

      Agentic Profiling in AI Red Teaming helps gather all relevant context about a target endpoint such as its business use case, background, key capabilities, technical architecture, and other critical information. This is carried out by an autonomous agent probing the target endpoint with the right prompts. All information gathered through this exercise is presented as the target's profile and is used downstream in AI Red Teaming scans.

8. Select Submit.

   Once the target is created, you can start a scan or view previously created targets.