I've written that library because I saw Vercel’s Chat SDK, but when I wanted to try it, I realized that it's not an SDK; it's just a template, deeply tied to Next.js. This library can be used everywhere. (I hope)
Try it out:
npx sourcewizard@latest install react-ai-agent-chat-sdk
(2/2)
Over the two weekend, I've (vibe)coded a library for creating agentic chats. It's a full-stack library - it provides React UI components for the chat, tools, and a backend endpoint implementation based on Vercel AI SDK. (1/2)
JSON not supporting trailing commas is one of the largest software engineering mistakes.
Just found your website, such a gem!
Now 4 months later she's released, the judge questioned the legal basis of her visa revocation and the fact she was detained for such a long time. This story is not over yet - she's waiting for another trial over smuggling the embryos.
Wanted to share an update on a story I've been following:
In February a Russian scientist in Harvard was detained at the border because she didn't declare flog embryos, her visa was revoked and she was sent to ICE detention center.
Found a really great infographic website that shows the US budget in a scaled down perspective so the numbers would be actually easier to comprehend: www.debtinperspective.com
HOLY SHIT! This might be the coolest thing ever.
oh you're a full stack engineer? how many weeks does it take you to answer the "what happens when I type google into the address bar" interview question?
interesting link previews...
Check it out here: typeconf.dev/storage
We currently working on the supporting secret managers and mcp, so stay tuned!
Now you will be able to upload your config to the Storage and update it whenever you want! Moreover, we support versioning, so you can quickly rollback your config if anything breaks. And the best thing? It’s open-source and completely free!
We’re launching Typeconf Storage alpha!
What is it?
With Typeconf you can define configuration in Typescript which will be compiled to plain JSON. You could the generated file in your app, but it would mean that you have to redeploy it on every change, which is slow. 🧵
"Ah, I understand now!"
proceeds to generate the exact same broken code🤦
This is an unbelivable story and it requires a lot more attention now: gofund.me/833d8794
A Harvard scientist who fled Russia because of the war is now has her visa revoked and detained by ICE. She's at risk of being deported back to Russia. How could this be happening?
This is fun
pure cinema
Years go by, but The Night Watch by James Mickens is still one of the best articles I've read about systems engineering: www.usenix.org/system/files...
well, currently it shows the “vibes” that it can, but the progress doesn’t stop here
Think of it, creating apps was limited by people having access to arcane languages and complicated algorithms (aka mainframes with operators), and now everyone gets a personal tool that can produce the same things for everyone!
Wild thought: It looks like with AI we’re entering the era of Personal Coding akin to Personal Computers.
I’m getting more amazed by AI code generation startups! rork.app generates both web AND mobile apps right in the browser. So you can preview and iterate on the mobile app without having to interact the emulator!
Interesting to hear more about it!
It seems my opinion on this is a bit more optimistic, we'll see how it will really progress, sooner or rather later. I agree that it would take a lot of work to transform these cool demos to something substantial.
Exactly, and what I saw was poor, duplicated code that worked as of that time, which would definitely need a lot of support later. But even this is a huge achievement! What stops people from training LLMs to use better frameworks, run code deduplication, refactoring, this are the problems to solve.
Can you share which ones? Currently they look like toys, but I’ve seen non engineers generate a hefty amount of working code with them, so I think it won’t be long until you can’t dismiss them that easily…
