Skip to main content

​
Overview

A Custom Client Action allows your AI Agent to execute workflows directly in the client (e.g., browser). This is particularly useful for client-side data handling or UI updates. Note: Custom client actions will not work in Slack or the in-app chat window. Example Use Case: Detect the user’s current location and show nearby restaurants, or auto-fill a form with profile data stored in the browser.

​
How to Create a Custom Client Action

​
Step 1: Create a Custom Client Action

  1. Go to the Actions tab.
  2. Click List from the left-hand menu.
  3. Select Create Action. Server Action One
  4. Choose Client Action under Custom Action.
  5. Enter an Action Title in the pop-up window and click Create. Client Action Two
  6. You’ll now see the action setup screen, where you can configure the details.
Example: Create a client action called getLocation to detect the user’s browser location.

​
Step 2: Basic Settings

  1. Name - Enter a descriptive name.
    • Example: getLocation
  2. Description - Explain what the action does and when it should be used.
    • Example: Detects the user’s current location using the browser and returns the coordinates for nearby content.
This helps the AI Agent know when to trigger the action. Client Action Two

​
Step 3: Add Parameters (Optional)

Parameters allow you to collect input data that the client action may need.
  • Click Add Parameter.
  • Fill in the fields:
    • Name: radius
    • Description: Distance in kilometers around the user’s location to search for restaurants.
    • Type: Number
    • Required: ✔️
Parameters Ca

​
Step 4: Enable the Action

At the top of the screen, toggle the switch from Disable → Enable to activate the action. You can disable it later if needed. Client Action Three

​
Step 5: Register Tools in Your Client

Finally, register your Client Action in your website or app using the registerTools method. See the Developer Guide: registerTools() for details and examples.

​
Example Workflow

  1. User asks: “Find coffee shops within 2 km of me.”
  2. AI Agent triggers the getLocation Client Action.
  3. The action detects the user’s location using the browser.
  4. The agent calls a restaurant API with the location and radius.
  5. Response: “Here are 3 restaurants within 2 km of your current location.”
With this setup, your AI Agent becomes interactive and context-aware, providing results tailored to the user’s environment.
⌘I