Stateful agent runtime for Java applications.
Graph workflows · ReAct agents · Context engineering · Human-in-the-loop · Multi-agent orchestration
Documentation · Quick Start · Examples · 简体中文
ARGI stands for Agent Runtime and Graph Intelligence and is pronounced "AR-jee" (/ˈɑːr.dʒiː/). It is a framework for Java developers building agents, workflows, and multi-agent applications. Forked from Spring AI Alibaba, it provides context engineering, human-in-the-loop, graph workflows, and distributed Agent-to-Agent (A2A) collaboration.
- Agent orchestration:
ReactAgent,SequentialAgent,ParallelAgent,RoutingAgent, andLoopAgent. - Context engineering: context compaction and editing, call limits, tool retries, planning, and dynamic tool selection.
- Graph workflows: conditional routing, parallel execution, nested graphs, state persistence, and interruption recovery.
- Agent runtime foundation: provider-neutral agent orchestration, workflow state, and embeddable debugging support.
- Open integrations: a neutral ChatClient with Spring AI, LangChain4j, and Google ADK adapters, plus tool calling, MCP, A2A, and Nacos.
- Visual debugging: an embeddable Agent Chat UI for Spring Boot applications.
Requirements: JDK 17 or later and Maven 3.9.1 or later. Build the framework with its Maven Wrapper and run the standalone examples with your local Maven installation.
git clone --depth=1 https://github.com/agentic-ai-java/argi.git cd argi # Install the local development modules. ./mvnw -DskipTests install # Configure any supported model provider and run the chatbot example. cd .. git clone --depth=1 https://github.com/agentic-ai-java/argi-examples.git cd argi-examples mvn -f examples/chatbot/pom.xml spring-boot:run
The examples also require the matching Extensions BOM and provider starters. See the example setup. Examples using older Spring AI agent builders require the direct-replacement migration described in API and Adapters; that external repository is not migrated by this change.
Open http://localhost:8080/chatui/index.html. See the Quick Start for other model providers.
| Module | Description |
|---|---|
| API and Adapters | Framework-neutral ChatClient contracts and Spring AI, LangChain4j, and Google ADK adapters |
| Agent Framework | Agent development and multi-agent orchestration |
| Workflow Graph Core | Stateful graph execution, persistence, and workflow runtime |
| Runtime API | JDK-only identity, definition and execution-control contracts |
| Agent API | Engine-neutral agent invocation and shared interception contracts |
| Agent Core | Graph-independent ReAct execution loop and tool lifecycle |
| Workflow Core | Versioned declarative workflow definitions and independent execution |
| Runtime Core | Engine-neutral Run/Task data, codecs and reference storage |
| Studio | Visual debugging UI for agents |
| Sandbox | Optional isolated execution environment for tool calls, maintained in Extensions |
| Spring Boot Starters | Built-in graph nodes, graph observability, and three adapter auto-configurations |
| Extensions | Model and document contracts, A2A, Nacos, AgentScope, storage, and other optional integrations |
| Examples | Chatbot, multi-agent, graph engineering, and documentation examples |
| Benchmark | argi-benchmark |
- Overview
- Quick Start
- Agent Framework tutorials
- Workflow Graph Core Quick Start
- Examples
- Provider-specific examples
Read the contribution guide before submitting changes. Report problems and suggestions through GitHub Issues.
ARGI is available under the Apache License 2.0.