License GitHub Stars GitHub Forks Maven Central DingTalk
[Cookbook] [中文README] [Examples]
AgentScope Runtime Java
This is the Java implementation of AgentScope Runtime.
- Deployment Infrastructure: Built-in services for session management, memory, and sandbox environment control
- Sandboxed Tool Execution: Isolated sandboxes ensure safe tool execution without system compromise
- Developer Friendly: Simple deployment with powerful customization options
- Framework Agnostic: Not tied to any specific framework. Works seamlessly with popular open-source agent frameworks and custom implementations
- 🚧 Observability: Trace and visualize agent operations comprehensively (under development)
Join our community on DingTalk:
| DingTalk |
|---|
- Java 17 or higher
- Maven 3.6+
Add the following dependency to your pom.xml:
<dependency> <groupId>io.agentscope</groupId> <artifactId>spring-boot-starter-runtime-a2a</artifactId> <version>1.0.2</version> </dependency> <!-- Add AgentScope Agent adapter dependency --> <dependency> <groupId>io.agentscope</groupId> <artifactId>agentscope-runtime-agentscope</artifactId> <version>1.0.2</version> </dependency>
The following example demonstrates how to delegate a AgentScope ReactAgent using AgentScope Runtime. The complete source code can be found in the examples directory.
- Create Agent Handler
Create a custom agent handler by extending AgentScopeAgentHandler:
public class MyAgentScopeAgentHandler extends AgentScopeAgentHandler { @Override public Flux<io.agentscope.core.agent.Event> streamQuery(AgentRequest request, Object messages) { // Create Toolkit and register tools Toolkit toolkit = new Toolkit(); if (sandboxService != null) { Sandbox sandbox = sandboxService.connect( request.getUserId(), request.getSessionId(), BaseSandbox.class ); toolkit.registerTool(ToolkitInit.RunPythonCodeTool(sandbox)); } // Create ReActAgent with tools ReActAgent agent = ReActAgent.builder() .name("Friday") .toolkit(toolkit) .model(DashScopeChatModel.builder() .apiKey(System.getenv("AI_DASHSCOPE_API_KEY")) .modelName("qwen-max") .stream(true) .formatter(new DashScopeChatFormatter()) .build()) .build(); // Convert messages and stream agent responses // See examples/README.md for complete implementation return agent.stream(queryMessage, streamOptions); } }
- Initialize Services and Deploy
Configure the agent handler with required services and deploy using AgentApp:
// Create and configure the agent handler MyAgentScopeAgentHandler agentHandler = new MyAgentScopeAgentHandler(); agentHandler.setStateService(new InMemoryStateService()); agentHandler.setSessionHistoryService(new InMemorySessionHistoryService()); agentHandler.setMemoryService(new InMemoryMemoryService()); agentHandler.setSandboxService(new SandboxService( new SandboxManager(ManagerConfig.builder().build()) )); // Deploy using AgentApp AgentApp agentApp = new AgentApp(agentHandler); agentApp.run(8090); // Server will listen on port 8090
Note
You can also use Kubernetes or Alibaba FC platform AgentRun to execute sandbox tools. Please refer to the examples directory for more details.
AgentScope Runtime Java implementation can be easily integrated with any agent frameworks developed in Java. Currently supported frameworks include:
- AgentScope Java
- Spring AI Alibaba, Langchain4j and more coming soon...
We welcome contributions from the community! Here's how you can help:
- Use GitHub Issues to report bugs
- Include detailed reproduction steps
- Provide system information and relevant logs
- Discuss new ideas in GitHub Discussions
- Follow the feature request template
- Consider implementation feasibility
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
For detailed contributing guidelines, please see CONTRIBUTING.md.
This project is licensed under the Apache License 2.0. See the LICENSE file for details.