Skip to main content
Tools let your agent interact with external systems—checking inventory, booking appointments, processing payments, or anything your API can do.

​
How Tools Work

When a user asks a question that requires external data, your agent:
  1. Recognizes the need to call a tool
  2. Extracts relevant parameters from the conversation
  3. Calls your API endpoint
  4. Interprets the response
  5. Replies to the user

​
Types of Tools

Connect to any REST API. You define the endpoint, parameters, and how to interpret responses.Examples:
  • Order lookup
  • Appointment booking
  • Inventory check
  • CRM updates
Let your agent fetch and read web pages in real-time.Examples:
  • Check current pricing
  • Read latest blog posts
  • Verify stock on your website
Generate images based on user requests using AI.Examples:
  • Create custom graphics
  • Generate product mockups
Transfer the conversation to another agent or human.Examples:
  • Escalate to support
  • Transfer to sales
  • Route to specialist
Connect to Model Context Protocol servers for complex integrations.Examples:
  • Database access
  • File system operations
  • Custom business logic

​
Creating an API Tool

1

Open Your Agent

Go to your agent and navigate to the Tools section.
2

Add New Tool

Click Add Tool and select API Tool.
3

Define the Tool

Fill in the tool configuration:
FieldDescriptionExample
NameA clear, descriptive namecheck_order_status
DescriptionWhen should the agent use this tool?”Use this tool when a customer asks about their order status”
API SchemaOpenAPI-style schema defining the endpointSee below
4

Configure the Schema

Define your API endpoint using a JSON schema:
{
  "type": "object",
  "properties": {
    "endpoint": {
      "type": "string",
      "const": "https://api.yourstore.com/orders/{order_id}"
    },
    "method": {
      "type": "string",
      "const": "GET"
    },
    "parameters": {
      "type": "object",
      "properties": {
        "order_id": {
          "type": "string",
          "description": "The customer's order number"
        }
      },
      "required": ["order_id"]
    },
    "headers": {
      "type": "object",
      "properties": {
        "Authorization": {
          "type": "string",
          "const": "Bearer your-api-key"
        }
      }
    }
  }
}
5

Test the Tool

Use the test panel to verify the tool works:
  1. Enter sample parameter values
  2. Click Test
  3. Review the API response
6

Save and Enable

Toggle the tool to Active and save your agent.

​
Tool Best Practices

The description tells the AI when to use the tool. Be specific:Good:
“Use this tool when the customer wants to know the status of their order. Requires an order number in format ORD-XXXXX.”
Less effective:
“Order lookup”
Your API should return clear error messages. In your system prompt, add instructions for handling failures:
“If an order lookup fails, apologize and ask the customer to verify their order number. If it fails again, offer to connect them with support.”
  • Use API keys or OAuth tokens
  • Implement rate limiting
  • Validate input parameters
  • Never expose sensitive data in responses
Return only what the agent needs. Large, complex responses slow down the conversation and may confuse the AI.Good response:
{
  "order_id": "ORD-12345",
  "status": "shipped",
  "eta": "2024-12-28",
  "tracking_url": "https://..."
}

​
Creating a Website Tool

Website tools let your agent read web pages in real-time:
1

Add Website Tool

In the Tools section, click Add Tool → Website Tool.
2

Configure the Tool

FieldDescription
URLThe page to fetch
PromptInstructions for what to extract
3

Example Configuration

URL: https://yourstore.com/products/widget

Prompt: Extract the current price, availability status, 
and key features from this product page.

​
Creating an Image Generation Tool

1

Add Image Tool

Click Add Tool → Image Generation.
2

Configure the Tool

FieldDescription
NameTool name (e.g., create_logo)
DescriptionWhen to use it
Prompt TemplateBase prompt for image generation
Credit CostHow many credits this uses

​
Creating a Handoff Tool

Transfer conversations to other agents or humans:
1

Add Handoff Tool

Click Add Tool → Handoff.
2

Configure the Tool

FieldDescription
Child AgentThe agent to transfer to
DescriptionWhen to initiate the handoff
3

Example Use Cases

  • General agent → Billing specialist
  • Sales agent → Technical support
  • Bot → Human operator

​
Connecting MCP Servers

For advanced integrations, connect Model Context Protocol servers:
1

Add Connector

Go to Settings → Integrations → Add Connector.
2

Enter Server URL

Provide the MCP server URL:
https://your-mcp-server.com
3

Authenticate (if required)

Some MCP servers require authentication. Complete any OAuth flow or enter credentials.
4

Add to Agent

In your agent’s Tools section:
  1. Click Add Tool → MCP Connector
  2. Select the connector
  3. Choose which tools to enable (whitelist)

​
Tool Credits

Some tools cost credits to use:
FieldPurpose
Credit CostCredits deducted when tool is used
Configure credit costs per tool in the tool settings. Users with monetization enabled will have credits deducted automatically.

​
SDK Alternative

Manage tools programmatically: 
// Update agent with tools
const updatedAgent = await client.agents.update(agentId, {
  tools: [
    {
      name: 'check_order_status',
      description: 'Look up order status by order number',
      apiSchema: {
        type: 'object',
        properties: {
          endpoint: {
            type: 'string',
            const: 'https://api.yourstore.com/orders/{order_id}'
          },
          method: { type: 'string', const: 'GET' },
          parameters: {
            type: 'object',
            properties: {
              order_id: { type: 'string', description: 'Order number' }
            },
            required: ['order_id']
          }
        }
      },
      active: true,
      creditCost: 0
    }
  ]
});

​
What’s Next

Set Up Monetization

Charge credits for tool usage.

View Analytics

Monitor tool usage and performance.