Feat v2.0.0 Release Notes

Breaking Changes

License Upgrade: AGPL → Apache 2.0

This is a significant milestone for the Feat project. We have upgraded our open source license from AGPL to Apache 2.0, providing a more business-friendly environment for enterprises and developers.

Why the License Change?

Since its inception in December 2024, the Feat project has experienced rapid development and iteration. Initially, we chose the AGPL license to protect the integrity of the open source project. However, as the project matured, we recognized that the "viral" nature of AGPL posed significant legal risks for enterprise users, becoming a major barrier to Feat's widespread adoption.

Apache 2.0 has become the de facto standard for enterprise-grade open source projects. Leading projects like Spring Framework, Kubernetes, and TensorFlow all use this license, striking a good balance between protecting creators' rights and promoting ecosystem prosperity.

Benefits for Enterprise Users:

• Business-friendly: Free to use in commercial closed-source projects without open source obligations
• Patent grant: Clear patent usage rights, reducing intellectual property risks
• Technology selection: Lower legal approval barriers, faster decision-making
• Ecosystem participation: Free to customize and extend without contribution requirements

Strategic Significance:

• Commercial application expansion: Enterprises can confidently apply Feat in fintech, e-commerce platforms, IoT, and enterprise internal systems
• Ecosystem prosperity: Attract more contributors, promoting the development of third-party libraries, toolchains, and solutions
• Long-term planning: Clear the path for future technology roadmap implementation

---

Upgrade Guide

Upgrading from v1.x

v2.0.0 is fully compatible with v1.x. Simply update the version number:

<feat.version>2.0.0</feat.version>

License Change Notice

After upgrading to v2.0.0, the project automatically applies the Apache 2.0 license. Enterprise users can confidently use it in commercial projects.

---

Links

• Documentation: https://smartboot.tech/feat
• GitHub Repository: https://github.com/smartboot/feat
• Gitee Repository: https://gitee.com/smartboot/feat
• Issue Tracker: https://gitee.com/smartboot/feat/issues

Feat v1.5.2 Release Notes

🚀 What's New

🤖 Feat AI: Enhanced Hook Mechanism for Observable AI Agents

We've significantly upgraded the Hook interface with three new streaming response methods, making AI Agent reasoning processes visible, observable, and interactive.

New Hook Methods

1. onModelReasoning(String) - Capture the model's "inner monologue" (e.g., DeepSeek-R1's reasoning_content)
2. onAgentReasoning(String) - Capture the ReAct agent's step-by-step thinking process
3. onFinalAnswer(String) - Receive the final answer content

Why This Matters

When using reasoning models like DeepSeek-R1, users want to see AI's thought process without waiting for the complete response. v1.5.2 solves this by separating reasoning and final answers in real-time.

Quick Example

ReActAgent agent = new ReActAgent(opts -> {
    opts.hook(new Hook() {
        @Override
        public void onModelReasoning(String reasoning) {
            // Show deep thinking indicator
            System.out.println("💭 Deep reasoning: " + reasoning);
        }

        @Override
        public void onAgentReasoning(String reasoning) {
            // Show step-by-step analysis
            System.out.println("🤔 Analyzing: " + reasoning);
        }

        @Override
        public void onFinalAnswer(String answer) {
            // Display final result
            System.out.println("✅ Answer: " + answer);
        }
    });
});

🎨 New AI-Powered Website Design

The official website homepage has been completely redesigned with:

• Dynamic gradient backgrounds with flowing light orbs
• Terminal-style code showcase with live editor effects
• F-E-A-T feature cards highlighting:
  • Free - Open source freedom
  • Elastic - Auto-scaling capabilities
  • Advanced - Cutting-edge technology
  • Tiny - Lightweight and fast
• Fully responsive design from desktop to mobile

📦 Installation

Maven

<dependency>
    <groupId>tech.smartboot.feat</groupId>
    <artifactId>feat-core</artifactId>
    <version>1.5.2</version>
</dependency>
<dependency>
    <groupId>tech.smartboot.feat</groupId>
    <artifactId>feat-cloud-starter</artifactId>
    <version>1.5.2</version>
</dependency>
<dependency>
    <groupId>tech.smartboot.feat</groupId>
    <artifactId>feat-ai</artifactId>
    <version>1.5.2</version>
</dependency>

v1.5.0

版本特性

• Feat Core：

  • [client] 重构 GZIP 输入流实现，支持完整的头部解析和更先进的异步解码算法。
  • [client] 修复 SSE 事件数据中的换行符转义问题。
  • [server] 修复静态资源处理器中的路径解析、HEAD 请求、空指针异常和 404 页面返回问题。
  • [server] 优化静态资源处理器的类路径资源定位功能，支持配置静态资源路径。
  • [server] 修复 HTTP 协议解析中的缓冲区边界检查和字节缓冲区位置错误。
  • [core] 添加方法实例化 Bean 的支持。
  • [core] 优化字节树子节点获取逻辑。
  • [router] 支持花括号 {} 形式的路径参数匹配。
  • 移除过时的 DeflaterInputStream 引用并优化异常处理。
  • 升级 smart-socket 至 1.8.3。

• Feat AI：

  • 实现 ReAct Agent 智能体，支持系统提示词、工具执行钩子和异常处理。
  • AI 代理工具改为异步执行模式，优化执行流程和消息处理机制。
  • 新增流式响应错误处理机制和失败处理机制。
  • 新增 Kimi-K2.5-Instruct 模型支持。
  • 重构提示词模板接口设计，优化 ReAct 智能体提示模板结构。
  • 新增子代理任务委托功能。
  • 实现 AI Agent 动作解析功能。
  • 新增 WebReader 网页阅读器功能，支持读取和转换网页内容。

• Feat Cloud：

  • 新增 MyBatis 集成支持，包括配置序列化、编译期 XML 配置解析和代码生成。
  • 新增分布式会话管理功能，集成 Redisun 支持集群会话管理。
  • 重构会话管理架构，实现本地会话管理器和集群会话管理器。
  • 改进 AOT 编译逻辑，支持扩展序列化器（MyBatis 和 Redisun 集成）。
  • RequestMapping 支持 {} 花括号形式的路径参数匹配。
  • 新增静态资源路径配置功能。
  • 新增 JSONArray 支持和结构体参数处理。
  • 增强控制器序列化器的参数处理能力，支持复杂类型参数。
  • 修复 JSON 解析空值导致的异常问题。
  • 修复列表参数反序列化问题。
  • 新增 Docker 部署支持，包括 JRE 版本和原生镜像示例。
  • 添加 Fat Jar 打包支持。
  • AsyncResponse 的 complete 方法返回类型调整为 AsyncResponse。

升级指南

1. 升级版本后处理编译报错问题。
2. 若使用了 AsyncResponse@complete 方法，注意其返回类型已变更为 AsyncResponse。
3. 若需技术支持可通过 Gitee ISSUE 提交反馈。

兼容性说明

1. AsyncResponse@complete 方法返回类型从 void 修改为 AsyncResponse。
2. ApplicationContext 类已设为 final，构造函数访问权限调整。
3. Feat AI 模块中 Agent 相关接口存在多处重构，如 AgentOptions 构造方式、工具接口等。

发版详情：https://mp.weixin.qq.com/s/fi0EmKhN9GHbUtZ6FfK3LA