JManus - Open source general-purpose agent for Java developers

It is in this context that Spring AI Alibaba JManus came into being. It is a completely open source OpenManus implementation based entirely on the Java language.

The original intention of JManus is to "bring Java programmers closer to AI". We hope to build a general AI Agent framework/platform based on the Java language and Spring AI Alibaba, so that Java programmers can use Agents more easily to meet their work and daily life needs.

As a sub-project of the Spring AI Alibaba project, the entire project has nearly 3k stars on GitHub and is increasing rapidly. In terms of large language model composition, JManus can seamlessly integrate multiple top large models including Claude 3.5, Qwen3, etc., allowing developers to fully utilize the advantages of each model.

After several months of development, JManus  has become a relatively complete system with the following system features:

• The OpenManus  multi-  agent  framework is perfectly implemented  with  the deep integration of Spring AI  and  Java  . Developers can easily build and manage multiple agents, making the division of labor and collaboration of complex tasks efficient and controllable.
  
• Supports Agent configuration through a web interface.  Developers and operation and maintenance personnel do not need to modify the underlying code. They only need  to perform simple operations on the  intuitive Web  management interface to flexibly adjust the Agent  's parameters, models, and tools, greatly improving ease of use and operation and maintenance efficiency.
  
• JManus  seamlessly supports  the MCP ( Model Context Protocol ) protocol , which means that  the Agent  can not only call large language models locally or in the cloud, but also interact deeply with various external services, APIs , databases, etc., greatly expanding the application scenarios and capability boundaries.
  
• JManus  natively supports  the PLAN-ACT  mode , which enables  the Agent  to have complex reasoning, step-by-step execution and dynamic adjustment capabilities. It is suitable for high-level AI application scenarios such as multi-round dialogues, complex decisions, and automated processes.
  

These functional features together build JManus' powerful AI Agent development platform, hoping to enable Java developers to quickly build, expand and implement intelligent systems for actual business with a very low threshold.

• Download software source code

git clone https://github.com/alibaba/spring-ai-alibabacd spring-ai-alibaba/spring-ai-alibaba-jmanus

• Make sure you have JDK 17 or later installed

• Set DashScope API Key
  
  

export AI_DASHSCOPE_API_KEY=xxxx

• Run using IDE

• Run with Maven

After the program starts, a local  http  page will be opened automatically.

Enter in the input box at the bottom of the page: "Query Alibaba's latest stock price through Baidu and save the results to a local file in the user directory."

Click "Send" to experience the intelligent question-and-answer and tool-calling capabilities brought by multi-agent collaboration.

ClickInput boxNext toPlanning Mode

Then inDemand InputLocation input: " Query the latest stock price of Alibaba through Baidu and save the results to the local file in the user directory ."

selectGenerate a plan, you can see the detailed execution plan.AlibabaAll changed to $companyName, thenAppending additional parametersPartially added configuration $companyName = Baidu. 

Then you can clickExecution planRun the task.