PM Universe
A constellation of product management resources that visualizes frameworks, case studies, and tactics for easier exploration.
I’m a builder who loves transforming complex ideas into delightful digital experiences. From backend services to interactive data visualizations, I enjoy navigating the entire stack and delivering work that feels both meaningful and elegant.
When I’m not experimenting with new technologies, you can find me capturing the night sky through astrophotography or exploring the latest trends in human–computer interaction.
Projects
A snapshot of experiments, tools, and platforms I’m building to make product thinking, automation, and storytelling more tangible.
A curated lineup of core platforms and system capabilities, from workflow engines to visual experiences.
A constellation of product management resources that visualizes frameworks, case studies, and tactics for easier exploration.
A minimalist e-signature workflow focused on privacy, speed, and clarity for teams that need agreements without the friction.
A personal knowledge library that syncs local reading notes with the cloud while keeping discovery fast and delightful.
Automated policy tracking that watches regulatory updates and surfaces alerts tailored to the teams who need to react.
A scoring and reporting engine that simplifies performance reviews with transparent rubrics and actionable insights.
An interactive editor for crafting Mermaid diagrams with instant previews and sharing-friendly exports.
两个纯前端的小宇宙,欢迎点击进入体验它们的节奏与情绪。
Publications
聚焦模型上下文协议与跨模型集成的工程实践。
ISBN 9787122494450
《MCP工程实践:AI模型上下文协议与跨模型集成全攻略》是一本聚焦AI时代模型交互瓶颈与跨系统集成难题的专业指南,它系统地解析了模型上下文协议(MCP)的核心原理、技术架构与工程落地方法。本书开篇直击AI模型的“上下文鸿沟”——浅层记忆局限、推理不透明性、位置偏置及专业领域盲区等问题,阐明MCP作为通用标准的诞生背景与核心价值;随后深入解析 MCP 的架构三元组(Host/Client/Server)、双向通信机制与沙箱安全模型,揭示其如何通过资源(感知数据库)、工具(操作手臂)、提示(任务剧本)三大能力,构建“耳聪目明且安全可控”的智能交互体系。书中不仅对比了MCP与传统方案(如OpenAI函数调用、插件机制)的优劣,还通过开发环境配置、服务器/客户端构建、工具箱案例(如纳米AI、字节Trae等)的完整流程演示,手把手指导读者从零搭建到可落地的跨模型集成系统。最后,针对安全风险(权限控制、数据加密、模型幻觉),提供了从开发阶段前置设计到运行时审计的全链路实践建议。