Add a tool to an agent (Customer Conversation) (preview)

[This article is prerelease documentation and is subject to change.]

You can add tools to your agent from the Build tab. Tools let your agent perform actions, connect to external services, and retrieve real-time data during conversations.

Adding a tool to an agent powered by the Customer Conversation harness is similar to adding a tool to a custom agent powered by the standard harness. To learn more, see Add tools to custom agents.

Add an existing tool from the Build tab

  1. Open your agent in Copilot Studio, and select the Build tab.

  2. Select the primary agent's card to open its configuration panel. If you've added any sub-agents, select a sub-agent's card instead to configure that sub-agent.

  3. In the Tools section, select Add tool.

  4. In the Add a tool dialog, browse or search for the tool you want to add. Use the tabs to filter by tool type: System tools, Popular, Prompt, Model Context Protocol, Connector, REST API, or Workflows.

  5. Select the tool to view its details, including a description and the actions it provides.

  6. Select Add to add the tool to your agent.

Configure a tool after adding it

After adding a tool, you might need to:

  • Authenticate: Some tools require you to sign in or provide credentials before they can be used. Follow the authentication prompts when they appear.
  • Select actions: For connectors that provide multiple actions, you can select which specific actions to make available to your agent.
  • Review the description: Ensure the tool's description accurately reflects what it does. The agent's orchestration runtime uses the description to determine when to invoke the tool.

Tips for effective tool descriptions

Tip

The tool's name and description are critical for accurate tool invocation, so:

  • Use clear, specific names that describe the tool (for example, "Create support ticket" rather than "Ticket tool")
  • Write descriptions that explain when the tool should be used and what it returns
  • Avoid vague or overlapping descriptions between tools
  • Avoid similar names or descriptions across tools, skills, topics, and sub-agents. If two resources look alike to the orchestrator, it can't reliably tell them apart when deciding which one to invoke.