Skip to main content

agents.restaurant_finder.agent

A2UI Restaurant Finder Agent using pydantic-ai.

This agent provides restaurant search and booking functionality, generating A2UI protocol messages for rich UI rendering.

This implementation uses a hybrid approach:

  • The LLM agent interprets user queries and calls tools
  • The A2UI response is built programmatically from tool results

RestaurantDeps Objects​

class RestaurantDeps()

Dependencies for the restaurant agent.

create_restaurant_agent​

def create_restaurant_agent(base_url: str) -> Agent[RestaurantDeps, str]

Create a new restaurant agent instance with the given base URL.

run_restaurant_agent​

async def run_restaurant_agent(query: str,
                               base_url: str = "http://localhost:8765",
                               max_retries: int = 2) -> dict[str, Any]

Run the restaurant agent with a query.

Arguments:

  • query - User's query (e.g., "Top 5 Chinese restaurants in New York")
  • base_url - Base URL for static assets
  • max_retries - Maximum number of retries on failure

Returns:

Dict containing the agent response with A2UI messages

handle_a2ui_action​

async def handle_a2ui_action(
        action_id: str,
        context: dict[str, Any],
        base_url: str = "http://localhost:8765") -> dict[str, Any]

Handle an A2UI action (button click, form submission, etc.).

Arguments:

  • action_id - The action identifier (e.g., "book_restaurant", "submit_booking")
  • context - The action context with relevant data
  • base_url - Base URL for static assets

Returns:

Dict containing the agent response with A2UI messages