All Categories

Are Chinese Flash Models Really Faster? Overthinking, Measured in Real Sessions

2026-09-25T12:00:00+09:00

Introduction

I used to avoid any model outside OpenAI and Anthropic. But the free GLM-5.2 that Devin offered completely changed my impression of Chinese open-weight models.

GLM-5.2 finds problems fast, which makes it great for quickly producing scripts and small tools, and for rapid trial-and-error in real environments (for example, having it verify low-level webOS commands to quickly produce a skill). The craziest part: during the promo period, none of the usage counted against my quota. Insane value for money.

From Isometric to 3D: Rebuilding the FFTA Battle Board and the "Far Side of the Moon" Problem

2026-09-08T06:00:00+09:00

Introduction

This was the scene of my first encounter with FFTA as a kid: a tiny GBA game, just 16 MB, that somehow contained such a vast, fully realized world and such exquisitely detailed battle visuals. When Mewt opened that old book, the real-world town of St. Ivalice was quietly rewritten into the fantasy world of Ivalice. Through that long summer vacation, I felt like I was living inside it too — forming a clan, learning “Two Swords,” mixing and matching jobs and races, setting out on adventure after adventure. In those moments, I really was one of Ivalice’s own. That, to me, is the true “Final Fantasy”!

Wrong Git Account Again? Conditional Includes & Multi-Account Isolation for GitHub CLI / AI Agents

2026-07-24T18:00:00+09:00

Background

As a developer constantly jumping across multiple projects, I frequently run into a headache-inducing issue: committing code in a company repository with my personal email, or mistakenly using my work email in a personal repo. Seeing a personal avatar show up in the commit history of a corporate project is always embarrassing.

Recently, I made this mistake again and had already pushed it to the remote repository. This time, I decided to fix the root cause once and for all so this careless error never happens again.

Does Every Apple Silicon MacBook Suffer from This Memory Leak? The Pitfalls of macOS CursorUIViewService

2026-07-04T21:00:00+09:00

Background

It has been about a year since I used Gemini to troubleshoot a freezing issue on Windows. Over the past year, I’ve maintained my habit of casting a skeptical eye on software produced by tech giants. Recently, I ran into another annoying issue on macOS. After debugging with an AI agent and cross-referencing developer forums, I confirmed it’s an XPC transaction memory leak in CursorUIViewService on Apple Silicon, introduced back in macOS Sonoma.

Dissecting the Switch 2 GameChat Backend: Multi-Region SFU and DynamoDB Consistency Trade-offs

2026-06-26T20:30:00+09:00

Background

A public backend talk from Nintendo is a rare thing. At AWS Summit Japan 2026, session CDN221 “The Technology Behind GameChat” (「ゲームチャットを支える技術」) walked through the service design behind GameChat, a built-in feature of Nintendo Switch 2. The speakers came from the System Development Department at Nintendo Systems Co., Ltd. (ニンテンドーシステムズ株式会社). That affiliation itself is worth noting: Nintendo Systems is not Nintendo proper but a joint venture Nintendo and DeNA founded in 2023, mainly responsible for developing the systems behind Nintendo’s online services.

Behind TBS's Large-Scale Interactive Live Streaming: AWS Serverless Design — AWS Summit Japan 2026 Recap

2026-06-26T12:00:00+09:00

Background

At AWS Summit Japan 2026, the session “Behind TBS TV’s ‘LOVE it!’ Large-Scale Streaming and AWS Serverless Design” (「TBSテレビ『ラヴィット！』大規模配信の裏側と AWS サーバーレス設計」) featured Mr. Kameda, Technical Product Lead in the Future Technology Innovation Division of TBS Television’s Media Technology Bureau, walking through the technical design of Kustamie, an interactive streaming platform. The session focused on a two-way live streaming scenario at a scale of tens of thousands of users: deliver low-latency video and real-time interaction reliably while protecting the API, database, and backend services from concentrated traffic spikes.

Handing Off Claude Code / Codex Sessions Between Two Macs: Syncthing + session-index-viewer

2026-06-19T10:00:00+09:00

Background

The pain point is straightforward: I have two MacBooks (A and B), and I heavily use both Claude Code and Codex for work and side projects on both machines. I wanted to switch seamlessly between them and resume previous sessions at any time.

Since their actual session data is stored locally as indexable session files (while state is also tracked upstream), as long as the session files on both sides are synchronized, I can resume right where I left off from either computer. Solving the synchronization problem was the first step.

Ditching Portainer for Dockge: What the Abstraction Hides Is the Most Expensive Part

2026-05-29T10:00:00+09:00

Portainer Really Is Showing Its Age

I love tinkering with Raspberry Pis. I had a Pi 4B that I hacked on for years.

Before GPU prices went through the roof last year, I bought a Pi 5 online and had a blast with it. For the container layer, I used Portainer just like I did on the 4B.

Portainer is great when you’re starting out: it has a UI, clickable buttons, one-click stack deployment, and you don’t need to memorize docker compose commands just to change an environment variable. But the longer you use it, the more small annoyances keep recurring.

From ctxd to Claude Code Connectors: There Are No Solo Heroes in the AI Era

2026-05-24T20:00:00+09:00

ctxd was a small CLI tool I wrote to let Claude Code automatically pull context from Slack, GitHub PRs, Confluence, and Jira. It worked great and saved me substantial time. But a few months later, Claude Code Connectors officially absorbed 80% of its core value.

This experience led me to rethink: what is the true value of personal tooling projects in the AI era?

How ctxd Came to Life

Between late last year and early this year, I wrote three CLI tools in succession:

When AI Pushes Back: Alignment Philosophies of Claude vs. Codex Through a Misread Algorithm Problem

2026-05-22T22:00:00+09:00

Background

In January 2017, I visited Tokyo for the second time in my life on a tourist visa to take the graduate entrance exam for the Department of Creative Informatics at the University of Tokyo.

Creative Informatics was somewhat unusual because it had no written math exam—only on-machine programming. While preparing for the exam, I worked through many past papers, finding several genuinely intriguing problem designs (far more interesting than routine LeetCode grinding).

Options Trading Crash Course: How I Learned with AI

2025-11-06T12:00:00+09:00

But somehow, we’ve now rushed into “the exchange for insurance policies and deposits.” We don’t care at all whether apples actually get delivered in the fall — we’re just purely, frantically speculating on the prices of the “insurance contracts” and “deposit contracts” themselves.

⸻ Gemini

Introduction

US share trading is pay the cash, take the shares. I am learning options by telling an AI I know none of this and asking it to teach me.

Exporting PR Context! Introducing the Custom CLI Tool "pr-dump"

2025-09-18T12:00:00+09:00

Introduction

As a non-native speaker working in a Japanese development team, writing PR review comments is always a small challenge for me. It requires not only making technical judgments, but also expressing them accurately and appropriately in Japanese.

So I often ask AI to help write drafts, but having to manually copy and paste the PR context every time—the summary, comment threads, diff—one by one, was honestly quite tedious.

Ask a Googler: How to Think 10X, Not 10%

2025-07-29T15:25:26+09:00

Background

Since joining the company, I’ve sat through a few sessions of Grow with Google, the 10X Innovation Culture Program. They were mostly watching videos plus a bit of brainstorming, and I never felt like I got anything concrete I could actually apply. Honestly, it felt like a pure waste of dev time.

I figured it would probably just fizzle out like that. But just as I was thinking that, HR came through in a big way: they brought in actual Googlers to answer our questions in person, and told us to prepare questions in advance.

I Couldn't Take It Anymore! Troubleshooting a Chrome Freeze and Filing an Official Bug with Gemini

2025-07-12T05:25:26+09:00

Background

I have no idea when it started, but whenever I launched Chrome on my PC, it would mysteriously freeze for one to two minutes. If you’re willing to just sit and wait, it’s not the end of the world.

But one night, staring at that completely frozen Chrome window, I snapped. I had to fix this.

First, I updated to the latest version. Didn’t work.
I uninstalled and reinstalled it. Didn’t work.
I wiped every trace of browser cache, reinstalled, logged in, and it froze again.

Behind the Scenes of Cross-Platform Play in "Monster Hunter Wilds" — AWS Summit Japan 2025 Retrospective

2025-06-26T12:05:26+09:00

Background

Within just one month of release, Capcom’s Monster Hunter Wilds surpassed 10 million copies sold worldwide. As the franchise’s first title to support cross-platform play, it reached millions of concurrent connections at launch. How did players actually feel about it? From the open betas to the official launch, player feedback voiced plenty of frustration over PC performance, content pacing, and slow updates—yet multiplayer networking received overwhelmingly positive reviews.

How Did PlayStation™Network Voice Chat Migrate to Amazon EKS? — AWS Summit Japan 2025 Session Recap

2025-06-26T05:25:26+09:00

Background

Voice chat is an application that requires a “long-lived connection lifecycle,” and migrating this kind of service to Amazon EKS is notoriously difficult. Sony Interactive Entertainment’s (SIE) session on migrating the PSN voice chat servers stood out for its technical depth among a sea of watered-down talks (one automaker is still “moving to the cloud,” 50% done), and it left a strong impression on me, bringing back memories of my own years wrestling with Kubernetes. Anyway, on to the agenda.

JINS AI: Bringing Generative AI to Physical Retail — AWS Summit Japan 2025 Retrospective

2025-06-25T12:00:26+09:00

Background

On Day 1 (June 25) of AWS Summit Japan 2025, during a session titled “アイウエア接客の未来を拓く～生成 AI で進化するあたらしい店舗体験への挑戦～” (Pioneering the Future of Eyewear Customer Service: Challenging New In-Store Experiences Evolving with Generative AI), eyewear retailer JINS shared the development journey, design details, and practical hurdles behind launching JINS AI.