Actions & Tools

The Actions tab lets you extend your agent with tools it can call during conversations.

HTTP Tools

Call external APIs during a conversation. The agent decides when to use a tool based on the conversation context.

Configuration

Field	Description
Name	Tool name the LLM sees (e.g., `check_order_status`)
Description	Tells the LLM when to use this tool
Method	HTTP method (GET, POST, PUT, DELETE)
URL	Endpoint URL. Use `{param}` for path parameters
Parameters	Input parameters with name, type, and required flag
Headers	Custom headers (e.g., Authorization)
Silent	If enabled, the agent won't speak while the tool runs

Example

A tool to check order status:

Name: check_order_status
Description: "Check the status of a customer's order by order ID"
Method: GET
URL: https://api.example.com/orders/{order_id}
Parameters: order_id (string, required)

Client Tools

Execute functions in the caller's browser (web calls only). Useful for UI actions like displaying information or navigating pages.

Field	Description
Name	Function name
Description	When to trigger this action
Parameters	Input data from the conversation
Preview Response	Mock response for testing
Silent	Suppress agent speech during execution

MCP Servers

Connect to Model Context Protocol servers for advanced integrations.

Field	Description
Name	Server identifier
URL	MCP server endpoint
Environment Variables	Key-value pairs passed to the server

Transfer Settings

Configure call transfers to humans or other agents.

Transfer to Human

Enable this to let the agent transfer calls to a human operator:

Destination Number — Phone number to dial
Trigger Description — When the LLM should initiate a transfer

Transfer to Other Agents

Select other agents that this agent can hand off to. The LLM will decide when a different agent is more appropriate based on the conversation context.

On this page