Feel the same, building regular shit is very boring. The fun is in the discovery, when you build dumb shit you actually get to do that. So many people find pulling from existing popular ideas to be creation, but it's not, it's curation.
This is the unfortunate truth. Also Bluesky is shrinking in users so coming here is not an obvious choice.
Markdown is markdown, and if its useful for #ai it should be useful for a human too
Use chisel docs with @astro Starlight and you have one set of markdown for your public docs, local TUI for devs, and a CLI for #llms to fuzzy search for context
pssst — if you like the idea of tv for developers, today is a great day to become a supporter (you get early access to new content + benefits/discounts on some big stuff we haven't announced yet 👀)
Congrats!
Guess I'll have to try out Threads more.
Clearly better than Wemby today too
January gets a bad rap but I don't think it deserves it. You do pretty much get off from all social expectations for the whole month.
I'm watching physical media like a hipster.
Astro Together: April 9, 2026 NEXT WEEK A London skyline showing tower bridge against a peachy sunset.
London. Next week. The Astro core team, more than 100 Astro devs, and… you?
Unique swag, talks from @mk.gg, @fancypenguin.party, @florian-lefebvre.dev, and more.
Sign up: luma.com/lzh944vx
Does he cry when he sees compilation times of modern compilers?
Opus has a 1 million token context window. For comparison, this is more than all of Astro's core package.
Conventional wisdom says application architecture doesn't change though. Make that make sense.
Isn't how we build apps today influenced by our wimpy short term memory?
This is genuinely awesome. The CMS landscape needed this kind of reset for a long time.
Crazy crazy day. My first Cloudflare blog, and it's a big one.
We're rebuilding WordPress as if it were built today. It's end to end TypeScript, works as an Astro plugin, and has secure plugin execution in dynamic workers.
It's called EmDash, try it now ⤵️
blog.cloudflare.com/emdash-wordp...
The first @astro.build v6 + Svelte 5 website I’m happy to ship to production. Prelim. result of migrating a „historically grown“ content-heavy website (500+ routes). We tried to tackle to complexity by nesting & some shallow routing trickery I owe to @vtbag.dev.
Really grateful for Server Islands 🙏
I think I might build a framework that's mostly just a skill that does imperative JavaScript using classes.
Along with the blog post there's a recording of the interview. Listen to the full interview OR the last 10 minutes (linked).
www.youtube.com/watch?v=ZbWx...
I checked and Astro is about ~700k tokens, so today's AI could fit it all into its short term memory and I certainly cannot come close to doing so.
I don't know *exactly* what this means, but I do think that the conventional wisdom of "what's good for humans is good for AI" is outdated.
Humans have better ability to access *long term* memory though, but it's still not easy. I have to reread areas of Astro code all of the time, I don't just recollect them when I need them.
A medium size codebase is like 20k LOC, so *maybe* 200K tokens. Can you fit an entire medium size codebase in your short term memory? I don't think many people can.
Bluesky has found a use for AI that empowers regular people to have more control over social media online. This is a good thing. Get over your weird reactionary revulsion to anything AI.
👏 @analogjs.org 2.4 for @angular.dev is out!
⚡️ Support for @vite.dev 8 w/Rolldown
🧪 Enhanced testing DX with @vitest.dev
🧑🚀 Improved integration with @astro.build
And more! 👇
dev.to/analogjs/ana...
Who will be the 100th registration for Astro Together London next week? The schedule is packed with excellent lightning demos from the core team, but it's the hangout time that I'm most looking forward to. It'll be a lot of fun. luma.com/lzh944vx
This is completely false, there are improvements all the time. This was just a week ago: research.google/blog/turboqu...
Because tailoring your software to luddites will simply ensure you're ultimately outcompeted and irrelevant.
No shade, just answering the question.
Opus has a 1 million token context window. A year ago it was 200k. About 3 years ago, 4k.
We're already way beyond human constraints and we're just getting started. So any future paradigms must account for this reality.
I could be wrong, but I think declarative is about human context constraints.
Should have trimmed your dependencies James
The replies to this finally made it clear to me that most people angry about AI don’t have the slightest clue as to what it is.
