AWorld

Build, evaluate and run General Multi-Agent Assistance with ease

Through AWorld (short for Agent World), you can quickly build real-world scenarios or task automation into agentic prototypes, then extend them into a generic agent or a team of agents to assist your real needs, like Manus.

Hope AWorld would bridge the gap between theoretical MAS (Multi-Agent System) capabilities and practical implementation in real-world applications and guide you into the AGI World. GLHF! 🚀

Core concepts:

• agent: AI-powered components that autonomously make decisions, use tools, do collaboration, and so on.
• swarm: define the topology structure of a multiple agents system.
• environment: the runtime supporting communication among agents and tools.
• task: complete runnable specific work that includes dataset, agents, environment, eval metrics, etc.
• client: submit various tasks for efficient execution.

Installation

With Python>=3.11:

python setup.py install

Environment Configuration

# Choose your preferred AI model(s) and set the corresponding API key(s)
# OpenAI (Required for GPT-3.5, GPT-4)
export OPENAI_API_KEY=sk-abcd1234wxyz5678...
# Anthropic Claude (Required for Claude 2, Claude 3)
export CLAUDE_API_KEY=sk-ant-api03xyz...

Usage

Running Pre-defined Agents (demo code)

Below are demonstration videos showcasing AWorld's capabilities across different agent configurations and environments.

Mode	Type	Demo
Single Agent	Browser use	▶️ Watch Browser Demo on YouTube
	Phone use	▶️ Watch Mobile Demo on YouTube
Multi Agent	Cooperative Teams	▶️ Watch Travel Demo on YouTube
	Competitive Teams	Coming Soon 🚀
	Mixed of both Teams	Coming Soon 🚀

or Creating Your Own Agents (Quick Start Tutorial)

Here is a multi-agent example of running a level2 task from the GAIA benchmark:

from aworld.agents.gaia.agent import PlanAgent, ExecuteAgent
from aworld.core.client import Client
from aworld.core.common import Tools
from aworld.core.swarm import Swarm
from aworld.core.task import Task
from aworld.config.conf import AgentConfig, TaskConfig
from aworld.dataset.mock import mock_dataset

import os
# Need OPENAI_API_KEY
os.environ['OPENAI_API_KEY'] = "your key"
# Optional endpoint settings, default `https://api.openai.com/v1`
# os.environ['OPENAI_ENDPOINT'] = "https://api.openai.com/v1"

# Initialize client
client = Client()

# One sample for example
test_sample = mock_dataset("gaia")

# Create agents
agent_config = AgentConfig(
    llm_provider="openai",
    llm_model_name="gpt-4o",
)
agent1 = PlanAgent(conf=agent_config)
agent2 = ExecuteAgent(conf=agent_config, tool_names=[Tools.DOCUMENT_ANALYSIS.value])

# Create swarm for multi-agents
# define (head_node, tail_node) edge in the topology graph
# NOTE: the correct order is necessary
swarm = Swarm((agent1, agent2))

# Define a task
task = Task(input=test_sample, swarm=swarm, conf=TaskConfig())

# Run task
result = client.submit(task=[task])

print(f"Task completed: {result['success']}")
print(f"Time cost: {result['time_cost']}")
print(f"Task Answer: {result['task_0']['answer']}")

Task completed: True
Time cost: 26.431413888931274
Task Answer: Time-Parking 2: Parallel Universe

Framework Architecture

AWorld uses a client-server architecture with three main components:

1. Client-Server Architecture: Similar to ray, this architecture:

   • Decouples agents and environments for better scalability and flexibility
   • Provides a unified interaction protocol for all agent-environment interactions

2. Agent/Actor:

   • Encapsulates system prompts, tools, and models with the capability to hand off execution to other agents
   • Agent fields and properties:

   Field	Type	Description
   id	string	Unique identifier for the agent
   name	string	Name of the agent
   model_name	string	LLM model name of the agent
   _llm	object	LLM model instance based on model_name (e.g., "gpt-4", "claude-3")
   conf	BaseModel	Configuration inheriting from pydantic BaseModel
   dict_conf	dict	Dictionary-structured configuration for safe key access
   trajectory	object	Memory for maintaining context across interactions
   tool_names	list	List of tools the agent can use
   handoffs	list	List of other agents this agent can delegate tasks to
   finished	bool	Flag indicating whether the agent has completed its task

3. Environment/World Model: Various tools and models in the environment

   • Computer interfaces (browser, shell, functions, etc.)
   • World Model

   Tools	Description
   browser	Controls web browsers for navigation, form filling, and interaction with web pages
   android	Manages Android device simulation for mobile app testing and automation
   shell	Executes shell commands for file operations and system interactions
   code	Runs code snippets in various languages for data processing and automation
   search	Performs web searches and returns structured results for information gathering and summary
   document	Handles file operations including reading, writing, and managing directories

Dual Purpose Framework

AWorld serves two complementary purposes:

Agent Evaluation

Standardized benchmarking of agent capabilities under a unified protocol:

• Unified task definitions to run both customized and public benchmarks
• Efficient and stable execution environment
• Detailed test reports measuring efficiency (steps to completion), completion rates, and token costs

Model Training

Continuous improvement through a collaborative competition cycle:

• Agent models improve to overcome challenges
• World models (environments) evolve to present new, more complex scenarios

Key Features

• 🌐 Environment Multi-Tool Support:

  • Browsers (Chrome, Firefox)
  • Android device simulation
  • Shell, code (Python), and apis (e.g., google_search)
  • File system (writing, managing on going)
  • Cloud sandbox for quick and stable deployment
  • Env as reward model

• 🤖 AI-Powered Agents:

  • Agent initialization
  • Delegation between multiple agents
  • Asynchronous delegation
  • Human delegation (e.g., for password entry)
  • Pre-deployed open source LLMs powered by state-of-the-art inference frameworks

• 🔄 Standardized Protocol:

  • Client-server protocol compatible with Model Contest Protocol (MCP)
  • Environment interfaces following gymnasium standards
  • Custom agent-environment protocol

• 🎛️ Web Interface:

  • UI for execution visualization
  • Server configuration dashboard
  • Real-time monitoring tools
  • Performance reporting

• 🧠 Benchmarks and Samples:

  • Support standardized benchmarks by default, e.g., GAIA, WebArena
  • Support customized benchmarks
  • Support generating training samples

Contributing

We warmly welcome developers to join us in building and improving AWorld! Whether you're interested in enhancing the framework, fixing bugs, or adding new features, your contributions are valuable to us.

For academic citations or wish to contact us, please use the following BibTeX entry:

@software{aworld2025,
  author = {Agent Team at Ant Group},
  title = {AWorld: A Unified Agent Playground for Computer and Phone Use Tasks},
  year = {2025},
  url = {https://github.com/inclusionAI/AWorld},
  version = {0.1.0},
  publisher = {GitHub},
  email = {chenyi.zcy at antgroup.com}
}

License

This project is licensed under the MIT License - see the LICENSE file for details.