
I can use my laptop keyboard for ~an hour before I start to feel strain these days The biggest loss is the ability to work casually from coffee shops 😓

Principal engineer @ Cloudflare Observability More active on BlueSky

I can use my laptop keyboard for ~an hour before I start to feel strain these days The biggest loss is the ability to work casually from coffee shops 😓

I was perfectly happy with the Apple keyboards until ~38 when they suddenly caused me no end of pain Now my keyboard looks like this, and my wrists feel good again
Best keyboard you can buy right now:

I just reduced storage, memory, and IO by ~75% and made queries ~85% faster on a database I’m working on. For me, the most rewarding use of AI for programming is performance optimization. In a few days I can do work that used to take months or years: write benchmarks, manipulate variables, rewrite code, goal-seek, make comparisons. The process is creative and the outcomes are concrete.

Just released a new Databases Internals episode on The GeekNarrator podcast It was a pleasure to have Andrew Lamb @andrewlamb1111 back on the show after 2 years. youtu.be/LBSfavhN4pw please like and repost for the love of open source and databases. 🙏

Probably the most asked for feature ever, spend limits for Workers and Durable Objects are now generally available on @Cloudflare No more billing surprises! Super proud of this and how well it works. Support for the rest of our products on Cloudflare is coming next. blog.cloudflare.com/spend-ma…

Talking to some peeps this morning, I think I've arrived at: agent traces should be a different Thing than the rest of your distributed traces, captured in parallel and treated differently. OTel can just be a transport layer. Capture the conversation state directly if you can.
I'm a big fan of otel-all-the-things, but trying to shove full LLM agent conversations into spans seems like using a signal that happens to be convenient?

I care about making otel-desktop-viewer pleasant to use, even more now that agents can find answers for us. There’s still more to learn by poking around the charts and data. Here’s some of the UI work I’ve been doing: ctrlspice.codes/viewer-ui-im…

If someone tried to attach an HTTP response body to a span, we'd immediately recognize that as a Bad Idea™️ but when it's an LLM response it's fine?

I'm a big fan of otel-all-the-things, but trying to shove full LLM agent conversations into spans seems like using a signal that happens to be convenient?

the most import skill in 2026 is still being able to tell your agents "wtf is this crap, no, not like this" if you stop doing that, you're basically fucked, despite what some devfluencers would like you to think.

Otel-desktop-viewer's been growing faster than I've been telling people about it, so I'm writing several catch-up posts. But attributes that contain maps and lists are now formatted nicely, and search finds values nested inside them! 😌

When I started EmDash, my goal was for WordPress sites to be "agent-portable", not compatible. Posts like this shows it works better than I could have imagined. Pixel-perfect theme conversion, and all custom plugins ported.
OK I did it. Migrated WordPress to Emdash. goodbye $600 bluehost server I've had for 15 years. 307 posts 6.7gb photos 4 custom content types 5 custom page templates 15 custom fields 1 semi-complicated plugin converted and a handful of simple ones Emdash skills moved it over, migrated off stylus(!) and it's pixel perfect on the front end. The page speed score went from 52 to 75. The Admin is fantastic as well - no problem for a WP user. Nice and fast

There is still no substitute for caring. Every piece of software getting rapidly better is happening because a human cares. And happens to be using AI to move really fast.

there are 20 startups offering identical compute/sandboxes as a service ... and then there's hunter who's tackling the much harder and higher value problem of a distributed filesystem, and happens to have a differentiated sandbox fall out of his design "by accident"
Hunter Leath @jhleath Nov 3 Article Concurrent agents on one file system I think there's a lot of confusion out there about what a file system really is. There's lot of places where we can assign blame. People's familiarity with SQL databases (designed to provide


Cloudflare is finally shipping the Temporal API for real this time! We can finally drop Moment.js / date-fns in favor of this modern Web API, shedding few more bytes from our runtimes

holy shit
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…