Notes on design, technology, and life, told in depth while looking for quiet corners of the digital age.
A custom hook isn't about moving code to another file—it's about redesigning a piece of state logic into a stable interface. This article clarifies what makes a real custom hook and when you shouldn't extract one, from three angles: extraction timing, problem boundaries, and interface design.
useRef offers stable references, but it's both a powerful tool and a dangerous trap. When is it a legitimate escape hatch from React's declarative model, and when does it reveal a code smell in your architecture? This article draws the line clearly.
When writing React, the easiest way to bloat your components isn't network requests or animations — it's storing just a little too much state. In this article, we'll start with the problem of derived state and gradually unpack the design issues and their solutions.
In React development, you often run into the problem of logging stale values inside closure functions. These issues appear to be React pitfalls at first glance, but they are actually the combined effect of two things: JavaScript closures and React's snapshot-based rendering.
useEffect was originally an escape hatch from React for synchronizing with external systems. But many treat it as the command center for all internal component logic. Remember this simple rule: if no external system is involved, you usually don't need an Effect.
Over time, many developers find that their experience with React Hooks shifts from clean and refreshing to increasingly messy. In this article, I'll analyze the root causes of this chaos and provide practical guidance to keep your hooks in order.
Ponytail is an open-source AI plugin that trains your coding assistant to think like a lazy senior engineer—solving problems by first using standard libraries and existing capabilities, and cutting out all unnecessary code.
As the number of tools grows, relying solely on the Vercel AI SDK Tool layer starts to feel the strain. In this article, instead of rushing into MCP, we first lay out the problems you'll face with a pure Tool architecture.
Reasonix is a terminal AI coding agent designed specifically for the DeepSeek API. Rather than being a generic AI assistant wrapper, it deeply leverages DeepSeek's prefix caching mechanism with a cache-first architecture, minimizing the cost of repeated context to the extreme.
This article will introduce how to use Codex Subagents to develop a product from scratch. You will see how multiple agents collaborate to build a space shooter game that runs directly in the browser, covering the entire process from design documents, visual assets, to testing and validation.
