Monitor and operate your Agent Crew

After deploying an Agent Crew in ZBrain Builder, three dedicated interfaces help you manage, monitor, and integrate your crew:

• Agent dashboard: Review input queues, track task execution, and inspect outputs.

• Performance dashboard: Analyze session-level metrics, usage data, and log details.

• API integration panel: Access API endpoints, authentication tokens, and sample code to connect your crew with external systems.

Together, these interfaces provide full operational visibility and control over your multi-agent workflows—from monitoring to performance tuning and system integration.

Agent dashboard

Once you deploy your Agent Crew, you are redirected to the ‘Agent Dashboard’. This interface helps you monitor agent activity, test the crew, and analyze generated outputs.

Queue

Located at the top of the dashboard, this section allows you to add an input queue and view and manage all previous inputs that the agent has processed. It includes the following elements:

• Search bar: Look up past requests using keywords.

• Filter: Apply filters like processing, completed, failed and pending to narrow down specific types of requests.

• Upload: Manually upload files to add them to the input queue.

Input panel

This panel displays the input content associated with the selected task from the queue. It displays the uploaded files. It provides the following actions:

• Download the original input file

• View in full screen for better readability

Report panel

This area shows the crew-generated output for the selected task. It includes the final output generated by the Agent Crew (e.g., article text, summary). It provides the following options:

• Download the report

• Copy the content

• Dislike to provide feedback

• View full screen

Crew activity

Displays a sequence of how the crew performed the task. Each entry reflects a specific step, such as writing, reviewing, or publishing.

How to provide feedback?

To report an issue with the output generated by the Agent Crew, follow these steps:

1. Locate the output panel: On the right side of the Agent Dashboard, locate the output generated for the selected queue item.

2. Click the dislike button: Click the thumbs-down (👎) icon. This will open a feedback popup titled ‘Provide additional feedback.’

3. Select feedback reason(s): In the popup window, choose one or more predefined reasons that best describe your concern:

• Don’t like the style

• Not factually correct

• Response not updated

• Not following the company’s guidelines

• Others

4. Add comments (optional): Use the ‘Add Specific Details’ field to include additional comments or clarification. This step is optional, but it helps the team better understand the context of your feedback.

5. Submit: Click Submit to send your feedback. Your input helps improve agent responses and refine crew behavior for future tasks.

Performance dashboard

The performance dashboard allows you to monitor how an entire crew is performing over time. It gives visibility into session usage, efficiency, satisfaction metrics, and associated token costs.

How to access

Click the speed meter icon located below the agent dashboard icon. This opens the performance dashboard, which displays the overall performance data of the agent crew and details of individual sessions.

Dashboard overview

Once inside, you'll see the following key elements:

• Agent name: Displayed at the top, along with a status tag indicating how long the agent has been active (e.g., Active since 1 hour ago).

• Search by record name: Use the search field to filter records by their name.

• Date filter: Use the ‘Select Date(s)’ control to narrow performance data for specific time periods.

Overall crew performance

The overall crew performance is displayed at the top of the dashboard. It showcases the following:

• Utilized time: Total time the agent crew actively spent processing requests.

• Average session time: Average duration users interact with the agent per session.

• Satisfaction score: User satisfaction rating based on their experience with the agent in each session.

• Token used: Total number of LLM tokens used across sessions.

Individual session details

Each session record in the dashboard includes the following columns:

How to use the agent crew via API

Click the ‘Use’ icon (located below the performance dashboard icon) to access the API integration interface. This section allows you to connect the agent crew with external systems or products by using the provided API endpoint.

API endpoint details

• Agent URL (endpoint): Click the copy icon to copy the endpoint URL for use in your application.

• Authorization token (API key): To authorize requests, include your API key in the authorization header using the bearer scheme. You can generate and manage your API key by navigating to:

  • Settings → My Account

• Sample code snippets: ZBrain provides code snippets for quick implementation in multiple languages. Make sure to replace <API_Key> with your actual API key and update the file, input, and workerId parameters as needed.