LlamaAgents Client

Async HTTP client for interacting with deployed llama-agents-server instances.

Installation

pip install llama-agents-client

Quick Start

import asyncio
from llama_agents.client import WorkflowClient

async def main():
    client = WorkflowClient(base_url="http://localhost:8080")

    # Run a workflow asynchronously
    handler = await client.run_workflow_nowait("my_workflow")

    # Stream events as they are produced
    async for event in client.get_workflow_events(handler.handler_id):
        print(f"Event: {event.type} -> {event.value}")

    # Get the final result
    result = await client.get_handler(handler.handler_id)
    print(f"Result: {result.result} (status: {result.status})")

asyncio.run(main())

Features

Run workflows synchronously or asynchronously
Stream events in real-time as a workflow executes
Human-in-the-loop support via send_event for injecting events into running workflows
Bring your own httpx.AsyncClient for custom auth, headers, or transport