📬 Get the next edition in your inbox
One email a week with the article, the episode and the reads. No fluff.
Edition number 20. The definitive answer to where your auth token should live, Maxim Salnikov on why AI adoption is a change management problem, and two brand-new sections at the bottom: a book recommendation and a meme of the week.
New article: The most secure way to store auth tokens
Where should a login token live: localStorage, memory, cookies, somewhere else entirely?
Where is it more secure? And how to make it more secure based on what you chose.
There's a clear winner at the end, an architecture the OAuth working group ranks above everything else. Check it out.
Podcast: Maxim Salnikov on AI harnesses at scale
New episode of Señors @ Scale is out. Maxim Salnikov is an AI Dev Tools Solution Engineer at Microsoft, leading AI-native development enablement for over 100 Microsoft and GitHub enterprise customers in Norway.
We get into the two new job titles emerging inside dev teams, context engineers and agent ops, the "AI superheroes" anti-pattern and why it doesn't scale, managing agent skills like dependencies with APM, and how to become an AI-native engineer by moving from one-to-one chat to orchestration.
Watch the full episode on YouTube
Community reads
Engineering leadership
Engineering managers are paying the price of rising expectations by Diego Quiroga — the EM role is expanding on every front at once, coaching, technical depth, strategic influence and more. How do we handle this growing responsabilities as an EM?
How technical should an engineering manager be? by Gregor Ojstersek — assess what your specific team is missing and fill that gap, with practical ways to keep a technical baseline through code review, proof-of-concept work, and time-blocking.
AI and agents
What is "loop engineering?" by Gergely Orosz — the emerging practice of wiring agents into continuously running, goal-directed loops instead of prompting them by hand.
An agent that hunts bugs in my app while I sleep by Debbie O'Brien — one agent explores the real app with Playwright hunting for bugs, a second reproduces and fixes the confirmed ones.
You can write in LLMese, but you don't have to by Fabrizio Ferri-Benedetti — structured semantic rewrites cut token counts by 35–50% without losing answer quality.
DSLs enable reliable use of LLMs by Unmesh Joshi — a domain-specific language narrows the solution space enough that LLM output becomes dependable, and the DSL, not the original prompt, ends up as the durable source of truth.
Safe(r) deterministic AI code generation through DSL partitioning by Brad Murry — let AI generate fast inside a DSL layer of pre-vetted primitives while critical infrastructure stays human-reviewed.
Design and the web platform
Every frame perfect by Niki Tonsky — every frame of a transition should look intentional and explainable, not just the start and end states.
Designing with mustard by Anna E. Cook — a sharp, satirical take on AI design claims.
In defense of polyfills by Lea Verou — the problems people blame on polyfills are really failures of the standards process.
Tooling
My thoughts on the Bun Rust rewrite by Andrew Kelley — I love drama. This is Drama on steroids. And it contains my favorite insult of all time: Jarred was already writing slop well before he had access to LLMs. (And this is the revised mellow version, the first article was even more heated)
Book recommendation
I love reading, so I am going to start sharing my latest reads
The Key to Happiness: How to Find Purpose by Unlocking the Secrets of the World's Happiest People by Meik Wiking
Meik Wiking runs the Happiness Research Institute in Copenhagen and wrote the number one bestseller The Little Book of Hygge, which makes him roughly the closest thing the world has to a professional expert on being happy. Here he digs into why the Danes keep topping the world happiness rankings and turns the research into something you can actually use: how purpose, togetherness, trust, and small daily rituals add up to a good life.
An easy, warm read that pairs well with logging off at a reasonable hour.
Meme of the week
Conferences
ZurichJS
ZurichJS is still on my list this year. Strong lineup, well-run community, and the kind of single-track conference you can actually keep up with.
First 3 tickets get 25% off: BCN140726
For others use code REACTJSBARCELONA_10 for 10% off tickets.
CityJS Athens
CityJS lands in Athens this October. Adding it here because the CityJS events are consistently worth the trip.
That's it for this one.
If you're enjoying these, share the subscribe link with someone who'd get something out of it, or with your team on Slack:
📬 Get the next one in your inbox
New articles, podcast episodes, community reads, and what is shipping in the JS world. One email a week, no fluff.
💬 Share this edition
📰 Other recent editions
Hi, Rockstar, today we have Mark Allen on the unwritten rules of engineering management, CSS-Tricks going independent again, TanStack Redact shrinking your React bundle, and the book about a software company built around joy
Hi, Rockstar, today we have a four-minute clip of why your JSON API is not RESTful, Vercel making coding agents survive crashes, a Zod upgrade that freed a third of the CPUs, and Bill Walsh on leading a team by fixing how it works
Hi, Rockstar, today we have Carson Gross, the creator of HTMX, teaching us about hypermedia, how Anthropic moved its monorepo off git in five months, TanStack Start 1.0, and the book for your first 90 days in a new role
Hi, Rockstar, today we have Ryan Townsend from Cloudflare on web performance, a hackathon in Barcelona with a €10,000 prize, how claude.ai got 3x faster in two weeks, and a book on the scaling laws behind cities and companies