Five long reads people saved with Xtracticle this week — two lab notebooks on code-made motion design, a candid look at how the Claude Code team reviews AI-written code, a one-day life-reset protocol and a Chinese crash course on how AI actually works.
Tip: change x.com to xtracticle.com in any post link to open it here.
by Florence · @flxrnc
A lab notebook of three Opus 5.5 animations, each printed with its original prompt and every follow-up correction, arguing that the unglamorous middle part — the second and third prompt — is where studio-quality motion comes from.
by DAN KOE · @thedankoe
Dan Koe argues that resolutions fail because people change their actions instead of their identity, and closes with a day-long protocol for working out what you actually want before setting any goal.
by Addy Osmani · @addyosmani
Addy Osmani explains how the Claude Code team kept reviewing code while merge volume doubled: automated review runs before a human looks, human attention is saved for design-level questions, and reverting became fast and cheap.
by 龙井longjing · @long_jingai
A two-hour corporate AI training course rewritten as one long article: what large models are, how to write prompts, what agents are and how to build a first one, aimed at readers starting from zero.
by rari🈀 · @0xwhrrari
Rari describes a five-layer pipeline for making videos with Opus 5.5 — the model writes a frame function, a renderer draws it, and a review loop decides what ships — so the next clip can be produced without rebuilding the whole setup.
Every week we look at the X Articles readers saved with Xtracticle and pick the ones worth your time, by hand. We show only the title, the author and a one-line summary in our own words, and link to the original on X — the writing belongs to its authors. Know a great X Article? Suggest it on GitHub.