📄 中文摘要
每个 AI 代理框架都需要一个工具系统。Python 的 LangChain 使用 @tool 装饰器，LangGraph 有 ToolNode。在 Go 中，我们围绕一个干净的接口模式构建了我们的工具系统，结果是向生产代理添加新 AI 工具只需大约 15 行代码和 2 行注册代码。这一过程并非夸张，读者将在文章结束时成功添加一个真实的、可供 AI 代理调用的工具。文章详细介绍了工具接口的设计，包括 6 个方法，其中 2 个是免费的，展示了如何高效地实现这一功能。
在 15 行 Go 中添加新 AI 工具：OpenAI 兼容工具系统解析
📄 English Summary
#3 - Add a New AI Tool in 15 Lines of Go: The OpenAI-Compatible Tool System Explained
Every AI agent framework requires a tool system. Python's LangChain utilizes @tool decorators, while LangGraph features ToolNode. In Go, a clean interface pattern was developed for the tool system, allowing the addition of a new AI tool to a production agent in approximately 15 lines of code and 2 lines for registration. This is not an exaggeration; by the end of the article, readers will have successfully added a real, functioning tool that their AI agent can call. The article elaborates on the design of the tool interface, which includes 6 methods, 2 of which are free, demonstrating how to implement this functionality efficiently.