Hacker News Reader

jonnonzon Ternlight – 7 MB embedding model that runs in browser (WASM)

Same here, when the fans started up I got startled. However my bread toaster often scares me too

jonnonzon Show HN: I measured the half-life of 41,301 Show HN launches. It's 7 hours

Annnnnnnnd it’s dead haha

jonnonzon GitHub Merge Queue Silently Reverted Code

Whats going on over there, is clippy in charge of the release button?

jonnonzon S. Korea police arrest man over AI image of runaway wolf that misled authorities

This is how the future will look!

jonnonzon Meta tells staff it will cut 10% of jobs

What happened to the metaverse ?I suspect maybe wasting all the resource wasn’t a good idea

jonnonzon Incident with multple GitHub services

Well I suppose they are finding out if you lay off too many people the IP of how the system works goes out the door with them.

jonnonzon An LLM That Watches Your Logs and Kills Compromised Services at 3am

yes!

jonnonzon What if the browser built the UI for you?

I believe branding is still possible, though more of a capability and platform approach like Stripe or Shopify. Local LLMs will likely make it more feasible in the future as costs decrease.

jonnonzon What if the browser built the UI for you?

This is actually really innovative! I like it a lot

jonnonzon What if the browser built the UI for you?

Why not both? Look at stripe or shopify for example

jonnonzon LocalChannels

LocalChannels ->

I wanted to explore a few technical areas, particularly packaging with Homebrew and using Rust as a CLI tool. Additionally, I am eager to give back to a community that has brought me joy over the past 10 years.

So, to be frank (though that's not my name), I grew tired of paying for ngrok for a basic use case: routing public -> local tunneling during development, with a persistent configuration I could start and stop as required.

LocalChannels is about the millionth public to local tunnel tool posted here. It's not groundbreaking; it uses SSE to route the requests to your local machine.

The platform is limited to 1000 "tenants" to maintain performance, and up to three persistent channels per tenant - so once it's full, it's full. I hope someone finds it useful :)