12/12/2025
Introducing SimVi: a game AI prototype that combines natural language processing with behavior tree systems. Through custom DSL and OpenAI API, players can describe behaviors in natural language, and the system automatically generates and executes corresponding behavior tree logic.
Read more →
12/11/2025
Exploring how AI is changing development efficiency, enabling developers with clear logic and rich experience to complete work that previously required massive technical resources at minimal cost. Analyzing the digital transformation opportunities and challenges traditional industries face in this chaotic era, and how to effectively use AI tools to seize opportunities.
Read more →
12/5/2025
Exploring how to design a self-evolving Agent system that allows AI to automatically grow needed capabilities during task execution. From fixed foundation to dynamic evolution, from tool building to mini-agent encapsulation, complete analysis of six core components' design thinking.
Read more →
12/5/2025
An in-depth comparison of seven mainstream multi-agent frameworks (AutoGen, LangGraph, CrewAI, AgentMesh, AutoAgent, InfiAgent, MAS²), analyzing from multiple dimensions including architecture patterns, collaboration methods, and development approaches to help developers choose the most suitable framework based on actual needs.
Read more →
12/4/2025
In-depth analysis of TSMC's Q1 2025 financial performance, AI-driven growth momentum, advanced process technology development, and global expansion strategy. The report indicates that TSMC's core position in the AI field and technological advantages position it to maintain strong growth in the coming years.
Read more →
12/3/2025
Facing marketing teams' constant requests to modify push notification lists, data analysts are overwhelmed. This article introduces how to use Cursor Agent with Metabase MCP, allowing marketers to operate directly using natural language while maintaining full control over API KEY permissions.
Read more →
12/3/2025
Detailed explanation of how to use LLM and web search technologies to automatically find corresponding official website domains from product names. Covers core design logic including search strategy optimization, batch judgment, and error handling.
Read more →12/2/2025
Introducing the Precomputed Tags architecture used in VibeBlog, how to replace traditional dynamic search with pre-computed JSON indexes, achieving a high-performance tag system with zero runtime computation.
Read more →
12/2/2025
A detailed documentation of the MCP (Model Context Protocol) servers I installed when setting up the development environment for the VibeBlog project, including each server's functionality, purpose, real-world use cases, and related resource links.
Read more →
12/2/2025
Documenting the process of building a MCP Server for deapi.ai text-to-image API from scratch and successfully generating the first AI image.
Read more →
12/2/2025
Exploring the core limitations of AI-assisted development: the ambiguity of natural language leads to cumulative errors. Proposing the 'Verification Cost Principle': you can only build solidly within the scope where you can verify AI output correctness with minimal time cost. Using the VibeBlog project as an example to analyze which parts best fit the low verification cost principle.
Read more →