agentUniverse

What is agentUniverse?

agentUniverse is a multi-agent framework based on large language models. It provides flexible and easily extensible capabilities for building individual agents. The core of agentUniverse is a rich set of multi-agent collaborative pattern components (serving as a collaborative pattern factory), which allows agents to perform their respective duties and maximize their capabilities when solving problems in different fields; at the same time, agentUniverse focuses on the integration of domain experience, helping you smoothly integrate domain experience into the work of intelligent agents.🎉🎉🎉

🌈🌈🌈agentUniverse originates from the real-world financial business practices of AntGroup (https://github.com/antgroup), dedicated to assisting developers and enterprises in effortlessly constructing domain-expert-level intelligent agents that collaborate to accomplish tasks.

We look forward to your practice and communication and sharing of Patterns in different fields through the community. This framework has already placed many useful components that have been tested in real business scenarios in terms of multi-agent cooperation, and will continue to be enriched in the future. The pattern components that are currently open for use include:

PEER pattern component: This pattern uses agents with different responsibilities—Plan, Execute, Express, and Review—to break down complex problems into manageable steps, execute the steps in sequence, and iteratively improve based on feedback, enhancing the performance of reasoning and analysis tasks. Typical use cases: Event interpretation, industry analysis.
DOE pattern component: This pattern employs three agents—Data-fining, Opinion-inject, and Express—to improve the effectiveness of tasks that are data-intensive, require high computational precision, and incorporate expert opinions. Typical use cases: Financial report generation.

More patterns are coming soon...

The LLM model integration can be accomplished with simple configuration, currently agentUniverse supported models include:

|-|Vendors| Models | |:-----:|:--------:|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------:| |<img src="https://github.com/user-attachments/assets/b7b0f2ce-3250-4008-b6d7-4712a983deb9" height="25">|Qwen| qwen3 Series（qwen3-235b-a22b、qwen3-32b、qwen3-30b-a3b, etc.）、qwen2.5-72b-instruct、qwq-32b-preview、qwen-max、… | |<img src="https://github.com/user-attachments/assets/5a997feb-bef4-4e53-ac3e-d38221e5399c" height="25">|Deepseek| deepseek-r1、deepseek-v3、deepseek-r1-distill-qwen-32b、… | |<img src="https://github.com/user-attachments/assets/0b50e555-65e8-49b2-b725-f3f71ee7daed" height="25">|OpenAI| GPT-4o、GPT-4o mini、OpenAI o1、OpenAI o3-mini、… | |<img src="https://github.com/user-attachments/assets/60fe0a70-0b47-4ac7-9bc9-8e860732ace9" height="25">|Claude| claude 3.7 sonnet 、Claude 3.5 Sonnet、Claude 3 Opus、… | |<img src="https://github.com/user-attachments/assets/334c7f09-7eae-4a65-a70f-2e6531964224" height="25">|Gemini| Gemini 2.5 Pro、Gemini 2.0 Flash、Gemini 2.0 Flash Thinking、Gemini 1.5 Pro、… | |<img src="https://github.com/user-attachments/assets/8e41c73f-3103-4305-ad1f-56116ea55523" height="25">|Llama| llama3.3-70b-instruct、llama3.2-3b-instruct、llama3.2-1b-instruct、… | |<img src="https://github.com/user-attachments/assets/19d264c6-e499-4913-9d6d-314d392f2246" height="25">|KIMI| moonshot-v1-128k、moonshot-v1-32k、moonshot-v1-8k、… | |<img src="https://github.com/user-attachments/assets/79572d9a-29d5-4c0e-a336-ce3f8018fb05" height="25">|WenXin| ERNIE 4.5 Turbo、ERNIE 4.5、ERNIE 4.0 Turbo、ERNIE 4.0、ERNIE 3.5、… | |<img src="https://github.com/user-attachments/assets/abb5311e-4d70-4e9c-8fca-e5129ae912fc" height="25">|chatglm| chatglm3-6b、chatglm-6b-v2、… | |<img src="https://github.com/user-attachments/assets/fe265f24-4ea6-4ff2-9b50-58ab6706a5f5" height="25">|BaiChuan| baichuan2-turbo、baichuan2-13b-chat-v1、… | |<img src="https://github.com/user-attachments/assets/41ffe268-392f-4ab9-b42d-e30dbd70d66b" height="25">|Doubao| Doubao-pro-128k、Doubao-pro-32k、Doubao-lite-128k、… |

For example, to use deepseek model, you can simply set DEEPSEEK_API_KEY value in the custom_key.toml file, and set the llm_model name in the agent configuration file to 'default_deepseek_llm' and you're all set. For more information about llm configuration, please refer to switch-the-llm.

Quick Start
How to build an agent application
Setup the visual agentic workflow platform
Why use agentUniverse
Sample Apps
Documents
Support

Quick Start

Installation

Using pip:

pip install agentUniverse

Run the first example

Run your first example, and you can quickly experience the performance of the agents (or agent groups) built by agentUniverse through the tutorial.

Please refer to the document for detail steps: Run the first example 。

How to build an agent application

Standard Project Scaffolding

Setup the standard project: agentUniverse Standard Project

Quick Start for Building a Single Agent

You can learn how to quickly build a single agent by reading the Quick Guide to Build Single Agent. This will help you understand how to enhance your agent's capabilities using tools, knowledge bases, RAG technologies, and more. Additionally, you will grasp the basic application development processes for agents, including configuration, testing, optimizing, deployment, and performance evaluation.

Building Typical Multi-Agent Applications

You can further understand how to break down intelligent capabilities into multiple agents in complex task scenarios and enhance your task performance through collaboration by referring to the Building Typical Multi-Agent App chapter.

Creating and Using Agent Templates

You can learn how to create effective agent patterns into templates through the chapter Creating and Using Agent Templates. This will greatly enhance the efficiency of constructing subsequent agents and facilitate dissemination.

Using and Publishing MCP Servers

You can learn how to quickly use or publish MCP servers in agentUniverse framework by referring to How to Use MCP Servers and How to Publish MCP Servers.

Agent Application Observability

agentUniverse provides a standardized observability framework for agent applications. Based on the OpenTelemetry protocol, it enables comprehensive data collection and monitoring of critical components such as agents, LLMs, and tools, facilitating full lifecycle tracking of agents. You can learn how to utilize these observability capabilities in the Agent Application Observability section.

Common Tips and Advanced Techniques

You can learn more advanced techniques in the agent application building process through other documents in the Get_Start, such as how to add a memory module into the intelligent agent process and how to effectively manage prompts within the project.

Set up the visual agentic workflow platform

agentUniverse provides a visual canvas platform for creating agentic workflow. Follow these steps for a quick start:

Using pip

pip install magent-ui ruamel.yaml

One-click Run

Run product_application.py in sample_

AgentUniverse

Install / Use

README