Connect and configure agent tools

Tools let a Fast Line Pro agent use external capabilities (for example APIs or bundled integrations) according to the catalog available in your environment. This guide covers the Tools section on the agent card and the Publish step so changes apply to live conversations.

When Needed

  • You want the agent to call external data or services from the tools catalog.
  • You need to disable a tool temporarily without removing it from the list.
  • You must fill in tool parameters when the catalog exposes fields for that tool.

What's Important to Know

  • Available tools come from the Tools catalog in the Fast Line Pro module.
  • Adding, disabling, or removing tools first updates only the agent card in the UI.
  • To apply changes to real agent behavior in bots, click Publish at the top of the agent card.
  • When parameters exist, use actions such as Configure parameters or the Parameters & settings dialog — complete required fields before publishing.

Prerequisites

  • You are logged in with integrator permissions.
  • Fast Line Pro is open (Apps → Fast Line Pro) and you are on the relevant agent card.

Step-by-Step Instructions

  1. On the agent card, open the Tools section.
  2. Open Tools catalog.
  3. Find the tool you need and add it to the agent (items already connected are usually marked Connected).
  4. If shown, open Configure parameters or Parameters & settings and fill in the fields.
  5. In the connected tools list, use the enable/disable switch to control whether each tool is active.
  6. To remove a tool, use the remove action and confirm in the dialog.
  7. Click Publish to save the configuration on the Fast Line Pro side and apply it to subsequent agent calls from scenarios.

What Happens After

After a successful publish, the agent uses the updated tool configuration. If you skip publish, dialogs keep using the last published configuration.

How to Verify It Worked

  • The Tools section shows the expected list and on/off states.
  • After Publish, you see a success confirmation when the UI provides one.
  • Run a test message through your scenario to confirm the agent behaves as expected with the tools.

MCP servers

MCP servers let an agent use external services through the Model Context Protocol. They appear under the MCP tab of the Tools catalog, and you add their tools to the agent like any other tool.

Add your own MCP server:

  1. In the Tools catalog, click Add MCP server.
  2. Fill in:
    • Name — a clear display name.
    • Server URL — the MCP server address (https://…).
    • Server Identifier — a unique id: lowercase letters, numbers, _ or -, up to 24 characters.
    • Timeout / SSE read timeout — keep the defaults (30 / 300 seconds) unless the server requires otherwise.
    • Headers (optional) — for example an Authorization header if the server needs a token. Header support depends on the AI engine version: if it is unavailable, you will see a warning after adding the server — in that case use token-based authorization in the server URL (query parameter).
  3. Click Add. The connection is checked automatically and the server's tools are loaded; the server appears on the MCP tab — add its tools to the agent and Publish as usual.

If the connection check fails, the form stays open with a warning: fix the headers and click Add again with the same Server Identifier to retry. If the server requires additional authorization on the provider side (OAuth), contact your administrator — the tools appear after authorization.

Connection status: every MCP server in the catalog shows a status dot next to its name — green means the server is connected and authorized, grey means it is not. The refresh button next to it re-checks the connection and reloads the server's tools; the result is shown above the list.

Visibility: MCP servers are scoped to your instance — you see only the ones you add, and they are not shown to other instances.

If you get a message that the name, server URL, or Server Identifier already exists, choose different values and try again.