はてなブックマーク

June 7, 2026 Loop engineering is replacing yourself as the person who prompts the agent. You design the system that does it instead. A loop here can be thought of a recursive goal where you define a purpose and the AI iterates until complete. I believe this may be the future of how we work with coding agents. However, its still early, I’m skeptical and you absolutely have to be careful about token

May 16, 2026 Right now, it’s too easy to let AI write the code while you skip the learning. The bug gets fixed but your mental model doesn’t move. We are silently trading future capability for present-day speed, and the tools won’t force us to do otherwise. That part has to come from you. There’s a default loop most of us have settled into. You paste in a spec or error message. The model hands you

April 7, 2026 Tip: Figure out your personal ceiling for running multiple agents in parallel. More agents running doesn’t mean more of you available. The narrative around agentic engineering right now is still mostly about throughput and parallelism, but almost nobody’s talking about what it actually costs the human in the loop. You’re holding multiple problem contexts in your head at once, making

May 5, 2026 Cognitive offloading is delegating to the AI and still owning the answer. Cognitive surrender is when the AI’s output quietly becomes your output and there is nothing you feel is left to check. For software engineers the line between the two moves under your feet most days, and most of us are crossing it without noticing. There’s a term I heard yesterday that I wanted to discuss: cogni

May 3, 2026 A senior engineer’s job is mostly the parts that don’t show up in the diff. Specs. Tests. Reviews. Scope discipline. Refusing to ship what can’t be verified. AI coding agents skip those parts by default. Agent Skills is my attempt to make them not optional. The default behaviour of any AI coding agent is to take the shortest path to “done.” Ask for a feature and it writes the feature.

April 19, 2026 A coding agent is the model plus everything you build around it. Harness engineering treats that scaffolding as a real artifact, and it tightens every time the agent slips. Roughly: anytime you find an agent makes a mistake, you take the time to engineer a solution such that the agent never makes that mistake again. We’ve spent the last two years arguing about models. Which one is s

March 14, 2026 Comprehension debt is the hidden cost to human intelligence and memory resulting from excessive reliance on AI and automation. For engineers, it applies most to agentic engineering. There’s a cost that doesn’t show up in your velocity metrics when teams go deep on AI coding tools. Especially when its tedious to review all the code the AI generates. This cost accumulates steadily, an

The Factory Model: How Coding Agents Changed Software Engineering February 25, 2026 Something shifted recently with agentic engineering that “feels” like the level of abstraction changed again. Not the usual kind of shift where tools get marginally better and workflows gradually evolve. A step change. Developers who have been writing software for decades are describing it the same way: the center

February 4, 2026 A year ago, Andrej Karpathy coined “vibe coding” to describe a gleefully reckless way of programming: you prompt, hand the keyboard to an AI, accept everything it spits out, don’t read the diffs, iterate by pasting error messages back in. It was a great label for a real thing - building quick prototypes or MVPs on pure AI autopilot. The problem is that “vibe coding” has become a s

January 5, 2026 The software industry sits at a strange inflection point. AI coding has evolved from autocomplete on steroids to agents that can autonomously execute development tasks. The economic boom that fueled tech’s hiring spree has given way to an efficiency mandate: companies now often favor profitability over growth, experienced hires over fresh graduates, and smaller teams armed with bet

January 3, 2026 When I joined Google ~14 years ago, I thought the job was about writing great code. I was partly right. But the longer I’ve stayed, the more I’ve realized that the engineers who thrive aren’t necessarily the best programmers - they’re the ones who’ve figured out how to navigate everything around the code: the people, the politics, the alignment, the ambiguity. These lessons are wha

October 2, 2025 Core Web Vitals measure user experience by assessing a website’s performance. This write-up is a history of how Core Web Vitals came to be based on my recollections from our work on it at Google from 2014 onwards. The initiative saved Chrome users over 30,000 years waiting with businesses seeing revenue and user-engagement lifts through optimizing for it, inclusive of 2025. Looking

September 2, 2025 Opinions expressed are solely my own and do not express the views or opinions of my employer Introduction I still remember the fall of 2008 when Google launched Chrome - a quirky new browser with a comic book as its press release. As someone who’s spent a long time on the Chrome team, I’ve watched this project grow from a secret skunkworks to a browser used by billions. Chrome tu

July 16, 2023 “Stick to boring architecture for as long as possible, and spend the majority of your time, and resources, building something your customers are willing to pay for.” - Kelsey Hightower When we’re starting new projects, there’s an undeniable allure to the latest languages, frameworks, and libraries. As engineers, we are, by nature, attracted to novel solutions. However, it’s critical

React Server Components, Next.js App Router and examples May 6, 2023 There were recently a number of discussions in the React community around the state of Server Components, the Next.js App Router and the future of tooling and rendering approaches. Here’s my attempt at a brief summary of concepts and discussions for folks that want highlights. I also wanted to share (lower down) more “complete” A

Tip - Use fetchpriority=high to load your LCP hero image sooner August 14, 2022 Tip: Add fetchpriority="high" to your Largest Contentful Paint (LCP) image to get it to load sooner. Priority Hints sped up Etsy’s LCP by 4% with some sites seeing an improvement of up to 20-30% in their lab tests. In many cases, fetchpriority should lead to a nice boost for LCP. Code snippets Priority Hints let you ad

In "Software Engineering - The Soft Parts" Addy Osmani shares lessons from his first 10 years at Google on the "soft skills" that can help engineers become effective and scale their effectiveness. This guidance should help junior, mid-career and even senior developers move forward, deal with changing technology, and navigate building non-trivial systems. Today I'll share some of the software engin

November 1, 2020 It’s historically been difficult for web developers to understand how the JavaScript in their apps performs in a wide range of cirumstances on real user devices. They’ve had to instrument code with poorly performing profiling hooks that could slow down page execution, without having the ability to collect stack samples efficiently. This is where the proposed native JavaScript self

September 17, 2020 VSCode Debug Visualizer is a VSCode extension that allows you to visualize data structures in your editor. This can be useful for visualizing watched values during debugging. I’ve found the extension can be helpful to visualize plots, tables, arrays, histograms and trees. Once the extension is installed, open some relevant scripts then navigate to the command palette and go to D

🕹 Puppeteer is a Node library which provides a high-level API to control headless Chrome or Chromium over the DevTools Protocol. This guide has recipes for automating Web Performance measurement with Puppeteer. An accompanying GitHub repository for this write-up is also available. Get a DevTools performance trace for a page load Puppeteer API: tracing.start() const puppeteer = require('puppeteer'

May 21, 2020 I recently spoke to veteran UX designer Steven Hoober about designing for touch interfaces. Hoober’s “Design for Fingers, Touch, and People” encourages UI designers to: Place primary content and actions at the center of the screen Place secondary actions and tabs along the edges Place tertiary functions behind menus Research how users perform specific interactions. Grip, task, context

April 8, 2020 Today, we’ll look at measuring React component render performance with the React Profiler API, measuring interactions with React’s new experimental Interaction Tracing API and measuring custom metrics using the User Timing API. For demonstration purposes, we’ll be using a Movie queueing app. The React Profiler API The React Profiler API measures renders and the cost of rendering to h

April 6, 2019 In this post, we’ll look at the new loading attribute which brings native <img> and <iframe> lazy-loading to the web!. For the curious, here’s a sneak preview of it in action: <img src="celebration.jpg" loading="lazy" alt="..." /> <iframe src="video-player.html" loading="lazy"></iframe> We are hoping to ship support for loading in ~Chrome 75 and are working on a deep-dive of the feat

February 18, 2019 How browsers schedule and execute scripts can impact the performance of web pages. While techniques like <script defer>, <link rel=preload> (and others) influence script loading, knowing how browsers interpret them can also be helpful. Thanks to Kouhei Ueno, we now have an up to date summary of script scheduling in Chrome.