Introduction In the previous post, we analyzed the architectural design principles of Claude Code. In this post, we go one level deeper to dissect Harness Engineering — the design principles behin...

Introduction Claude Code is a CLI-based AI coding agent developed by Anthropic. It’s an agentic tool that directly reads, modifies, builds, and tests code from the terminal, and has evolved rapidl...

Introduction In the previous post, we covered the principles of Unity Job System and Burst Compiler. In Part 4, we explored the cache hierarchy, the basics of AoS vs SoA, and the relationship betw...

Introduction To drive thousands of agents at 60fps in Unity, a single main thread is not enough. Path sampling, separation steering, distance calculations, matrix transforms — all of these must be...

Introduction Multithreaded programming is an unavoidable topic in game development. As CPUs have evolved toward increasing core counts rather than clock speeds, single-threaded programs can no lon...

Introduction The ability of AI agents to learn and remember long-term beyond a single conversation is a central topic of AI research in 2025–2026. Claude Code is one of the most aggressively exper...

Introduction This article is the 6th installment of the CS Roadmap series. In Part 1, we saw the 100x difference between L1 cache (~1ns) and RAM (~100ns). We also saw that cache locality is t...

Introduction This article is the 5th installment of the CS Roadmap series. In Part 4, we saw how trees guarantee O(log n) through hierarchical structure. BSTs search while maintaining order, ...

Introduction Claude Code’s Skills system has received a major update. What was once a simple structure — writing instructions in a single SKILL.md file — has evolved into a complete skill developm...

Introduction This article is the 4th installment of the CS Roadmap series. In Part 3, we saw that hash tables’ O(1) doesn’t come for free. Hash function quality, collision resolution strategi...