Login
Updated on July 15, 2026

Agent Group Dashboard

Estimated reading: 8 minutes

The Agent Group Dashboard provides an overview of agent group activity, compliance, policy events, and model usage within a project.

The dashboard consists of:

  • Agent Group Overview
  • Agent Controls

Agent Group Overview

The Agent Group Overview displays operational and compliance metrics across all agent groups.

The overview includes:

  • Trace Activity & Compliance – Trace volume and compliance outcomes.
  • Per-Agent Breakdown – Activity metrics for individual agent groups.
  • Top Triggered Policies – Policies triggered most frequently during execution.
  • Recent High-Risk Events – Recent events flagged as high risk.
  • Model Usage – Model usage across agent groups.
  • Tool Invocations – Tool usage during agent execution.

Trace Activity & Compliance

Trace Activity

The Trace Activity chart displays the number of traces processed over the last seven days, grouped by compliance outcome.

Compliance outcomes include:

Status Description
Passed Trace completed successfully without any compliance violations.
Redacted Sensitive content was detected and masked during processing.
Blocked Trace was blocked due to one or more compliance violations.
Trace Activity by Compliance Outcome

Compliance Split

The Compliance Split widget displays the distribution of trace outcomes across all processed traces as a donut chart.

Trace outcomes are categorized into:

Status Description
Passed Traces completed without compliance violations.
Redacted Traces where sensitive content was masked or modified.
Blocked Traces prevented from completing due to compliance violations.
Compliance Split

Per-Agent Breakdown

The Per-Agent Breakdown widget displays compliance and performance metrics for each agent group.

Health Score represents the percentage of requests handled successfully from a compliance perspective.

Formula:

Health Score = (Passed Traces + Sanitized Traces) ÷ Total Traces

The following metrics are displayed for each agent group:

Metric Description
Traces Total number of requests processed.
Passed Requests completed without compliance violations.
Sanitized Requests where sensitive data was successfully masked or redacted.
Blocked Requests prevented from completing due to policy violations.
Compliance Health Percentage of compliant requests.
Avg Runtime Average processing time per request.
Policies Number of active policies applied to the agent group.
Per-Agent Breakdown

Configure Agent Group

The Configure Agent Group screen allows users to define the operational settings, resources, and constraints for an agent group. Through a guided workflow, users can configure agent behavior, assign tools and skills, apply compliance requirements, and set execution limits.

Step 1: Agent Information

Provide the basic information required for the agent group.

Field Description
Description Describe what the agent group is responsible for, what tasks it is intended to perform, and how it is expected to behave.

Step 2: Adapter

Select the adapter that the agent group will use during execution.

Field Description
Adapter Choose an adapter from the available list. The selected adapter determines how the agent group interacts with underlying services and execution environments.

Supported Adapters: LangChain, Google ADK, Crew AI, and Strands.

Step 3: Tools Configuration

Manage the tools available to the agent group.

  • Selected Tools — Choose the built-in and custom tools that agents in the group can use during execution.
    Refer to Tools for more information.
  • Blocked Tools — Restrict access to tools that are unnecessary, sensitive, or irrelevant to the agent's intended purpose.
  • Custom Tools — If a required tool is not available, click Add Custom Tool to create a new project-scoped tool and make it available for selection.
    Refer to Custom Tools for more information.

Step 4: Agent Group Limits

Set execution boundaries and resource usage limits to help manage agent performance and system consumption.

Setting Description
Max Agent Calls Specify the maximum number of agent invocations allowed during execution.
Timeout (sec) Specify the maximum execution time, in seconds, before the request is terminated.
Input Tokens Specify the maximum number of input tokens that can be processed in a single request.
Output Tokens Specify the maximum number of output tokens that can be generated during execution.

Step 5: Compliance

Assign compliance to ensure agent interactions and outputs follow defined governance and regulatory requirements.

  • Select one or more compliance configurations from the available list.
  • Applied compliance rules are enforced during agent execution.
  • The policies associated with the selected compliance are listed. If you do not want to include a policy, you can disable it.

Step 6: Skills

Select the skills that should be available to the agent group during execution.

  • Selected Skills — Choose from built-in or project-specific custom skills available within the platform.
    Refer to Skills for more information.
  • Custom Skills — Click Add Custom Skill to create and register a new project-scoped skill. Newly created skills become immediately available for selection.
    Refer to Custom Skills for more information.

Step 7: Export Agent Slug

Export the agent group's unique identifier for use in integrations, deployments, or external references.

  • Generate the agent slug after completing the configuration.
  • Use the exported slug to reference the agent group in external systems and workflows.

Save Configuration

After completing all steps, review your settings and save the agent group. Once saved, the configured agent group becomes available for use within the selected project.

Trace Investigation

The Trace Investigation page provides a detailed audit trail for each request processed by an agent.

Each trace includes:

  • Input Context – The user request submitted for processing.
  • Processing Details – Information about the agent, model, and tool execution.
  • Response Output – The final response generated by the agent.

Trace outcome indicators:

  • Green – Passed
  • Amber – Redacted
  • Red – Blocked
Trace Investigation

Individual Trace Investigation Details

The Individual Trace Investigation Details page provides a detailed view of a single trace.

1. Execution Timeline

Displays all execution steps in chronological order.

2. Transaction Details Panel

Displays the request input, intermediate processing, model and tool interactions, and the final response.

3. Compliance Policies Panel

Displays the policies applied during execution. Expand a policy to view its evaluation results and the actions taken.

Individual Trace Investigation Details

Top Triggered Policies

The Top Triggered Policies widget displays the compliance policies that are triggered most frequently across all agent groups.

Severity levels are indicated by color:

  • Orange – High priority
  • Red – Critical
Top Triggered Policies
Note: This widget is read-only.

Recent High-Risk Events

The Recent High-Risk Events widget displays the latest policy enforcement events across all agent groups.

Each event includes:

  • Policy – The policy that was triggered.
  • Risk Category – The type of compliance risk identified.
  • Agent Group – The agent group associated with the event.
  • Time of Occurrence – The date and time when the event occurred.

Severity levels are indicated by color:

  • Red – Critical
  • Orange – High priority
Recent High-Risk Events
Note: This widget is read-only.

Model Usage

The Model Usage widget displays the number of LLM calls made by each model across all agent groups.

Each model includes:

  • Model Name – The name of the LLM model.
  • LLM Calls – The total number of calls made to the model.
  • Usage Bar – A visual representation of the total number of LLM calls.
Model Usage
Note: This widget is read-only.

Tool Invocations

The Tool Invocations widget displays the number of times each tool was invoked across all agent groups.

Each tool includes:

  • Tool Name – The name of the tool.
  • Invocation Count – The total number of times the tool was invoked.
  • Usage Bar – A visual representation of the total number of invocations.
Tool Invocations
Note: This widget is read-only.

Agent Controls

The Agent Controls page allows administrators to manage the execution of agent groups in real time. It provides controls for managing all agent groups, individual agent groups, and bulk operations.

Overall Controls

The Overall Controls section provides actions that apply to all agent groups.

  • Start All – Starts processing new traces across all agent groups.
  • Emergency Stop All – Immediately stops processing across all agent groups, including requests that are currently in progress.

Per-Agent Controls

The Per-Agent Controls section provides actions for managing individual agent groups.

  • Start – Starts processing new traces.
  • Pause – Temporarily pauses new trace processing while keeping the agent group available.
  • Stop – Immediately stops the agent group, including any requests that are currently in progress.
  • Drain – Stops accepting new traces while allowing in-progress requests to complete.
  • Adjust Rate Limits – Sets the maximum number of traces the agent group can process per minute.

Danger Zone

The Danger Zone section provides bulk actions that affect all agent groups.

  • Pause All Agents – Temporarily pauses processing across all agent groups.
  • Drain All Running Agents – Stops accepting new traces while allowing in-progress requests to complete.
  • Stop All Agents – Immediately stops processing across all agent groups, including any requests that are currently in progress.
Agent Controls
Note: Use Drain to stop an agent after all current executions are completed. Use Stop to terminate the agent immediately, including any requests that are still in progress.
Agent Management - Previous Agent Group
Share this Doc

Agent Group Dashboard

Or copy link

CONTENTS

Powered By Sutherland

Robility Chatbot
Robility Assistant
Online