Explore JManus: A Java‑Native AI Agent Framework Powered by Spring AI Alibaba

Overview

JManus is a fully Java‑based, open‑source implementation of the OpenManus multi‑agent framework. It is a sub‑project of Spring AI Alibaba and integrates top‑tier large language models such as Claude 3.5 and Qwen 3, allowing Java developers to build AI agents for work, daily life, or business scenarios.

Key Features

OpenManus‑compatible multi‑agent framework – Deep integration of Spring AI and Java enables efficient creation, management, and collaboration of multiple agents.

Web UI configuration – Agents, models, and tool parameters can be adjusted through an intuitive web interface without code changes.

Model Context Protocol (MCP) support – Agents can invoke local or cloud LLMs and interact with external APIs, databases, and services.

Native PLAN‑ACT mode – Provides step‑by‑step reasoning, dynamic plan adjustment, and execution suitable for multi‑turn dialogue and complex automation tasks.

Prerequisites

JDK 17 or newer (verify with java -version)

DashScope API key for accessing Alibaba LLM services

Getting Started

Clone the repository, switch to the JManus module, configure the API key, and start the application.

git clone https://github.com/alibaba/spring-ai-alibaba
cd spring-ai-alibaba/spring-ai-alibaba-jmanus
export AI_DASHSCOPE_API_KEY=xxxx
mvn spring-boot:run

The application launches a local web page (e.g., http://localhost:8080) where you can enter natural‑language requests such as:

"Search Alibaba’s latest stock price via Baidu and save the result to a local file."

The UI displays a generated plan, allows you to edit variables (e.g., replace the company name with $companyName) or add extra arguments, and then execute the plan.

Example Workflow

Enter a request in the input box, e.g., "Query Alibaba’s latest stock price through Baidu and store the result in a local file."

Click the Generate Plan button – the system shows a detailed execution plan.

Modify placeholders (e.g., replace the hard‑coded company name with $companyName) and optionally add extra parameters such as $companyName=百度.

Click Execute Plan to run the task. The agents invoke the selected LLM, call external services, and write the result to the specified file.

Source Code

Project repository: https://github.com/alibaba/spring-ai-alibaba

JManus module path: https://github.com/alibaba/spring-ai-alibaba/tree/main/spring-ai-alibaba-jmanus