frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

U1.lol – a free link shortener, no account or ads

https://u1.lol
3•noisecache•46m ago

Outlines – Structured Outputs for LLMs

https://github.com/dottxt-ai/outlines
1•modinfo•33s ago•0 comments

On the painful prescience of Seinfeld

https://alexberenson.substack.com/p/on-the-painful-prescience-of-seinfeld
1•CGMthrowaway•57s ago•0 comments

Prediction and Entropy of Printed English

https://archive.org/details/bstj30-1-50
1•Betelbuddy•3m ago•0 comments

I built a Claude Code terminal on the NDS

https://github.com/logancarrillo/ds-claude-terminal
1•tbgtech•3m ago•0 comments

Taco Bell drive-thru AI bot increased employee retention by making jobs easier

https://fortune.com/2026/10/14/yum-brands-executive-data-center-hate-ai-tools-improvement/
1•TMWNN•5m ago•1 comments

Cynical people are often seen as smarter. Here's why that might not be true

https://www.washingtonpost.com/lifestyle/2026/10/16/cynical-people-are-often-seen-smarter-heres-w...
1•bcaulfield•7m ago•0 comments

Native, free and open-source WireGuard client for macOS

https://github.com/ventz/wombat-wireguard-mac-app
1•pkaeding•8m ago•0 comments

New Data Centers in NM and El Paso Could Undo Millions Spent on Climate Plans

https://elpasomatters.org/2026/10/14/el-paso-data-centers-meta-project-jupiter-climate-emissions/
1•hn_acker•8m ago•1 comments

Researchers Discover ChatGPT Can Drive a Car. Grok, on the Other Hand

https://www.thedrive.com/news/uh-oh-chatgpt-can-drive-a-car-and-grok-crashed-immediately
2•speckx•8m ago•0 comments

The Fatal Flaw of 911

https://www.themarshallproject.org/2026/10/16/911-book-policing-mental-health
6•plunge•10m ago•1 comments

Woman faces changing 211 accounts after .name domain termination

https://domainnamewire.com/2026/10/16/woman-faces-changing-211-accounts-after-name-domain-termina...
2•speckx•14m ago•0 comments

Infinite Jeffs

https://www.amazon.com/dp/B0GTJ4P1PT
4•gregsadetsky•15m ago•1 comments

Google's Core Web Vitals with Soft Navigations, Reflecting SPA Performance

https://web.dev/blog/crux-experimental-soft-navs
1•powdercloud•16m ago•0 comments

Load-Bearing

https://pluralistic.net/2026/10/15/sovereign-it/
2•hn_acker•16m ago•0 comments

Mathematical Foundations of Generative AI, IISc Bangalore

https://nptel.ac.in/courses/106108004
1•yarapavan•16m ago•0 comments

Show HN: Make house walkthrough videos only from photos – TensorPix

https://tensorpix.ai/usecases/real-estate-video-maker
1•barty777•17m ago•0 comments

Trump Bought $5M in SpaceX Debt Two Days Before Space Transport Executive Order

https://www.techdirt.com/2026/10/16/trump-acquired-5-million-in-spacex-debt-two-days-before-space...
6•hn_acker•18m ago•1 comments

Show HN: Donut Browser, the most private anti-detect browser

https://github.com/donutbrowser/donutbrowser
2•andrewzeno•19m ago•1 comments

Adapting Our Processes to the Changing World of Scientific Publication

https://blog.tmlr.org/2026/adapting-our-processes-to-the-changing-world-of-scientific-publication/
1•theanonymousone•20m ago•0 comments

A Trip Down Market Street (1906)

https://en.wikipedia.org/wiki/A_Trip_Down_Market_Street
1•Alexsky2•20m ago•1 comments

Scaling Pre-Training on Super Smash Bros. Melee

https://ericyuegu.com/hillclimbing-melee/
2•Samin100•23m ago•0 comments

NASA's Most Off Limits Room

https://www.youtube.com/watch?v=89zdHsb2pWU
2•Betelbuddy•25m ago•0 comments

How to Avoid AI Brain Rot

https://twitter.com/MilesCranmer/status/2111105654211608847
3•madars•25m ago•1 comments

"Traditional, but Puzzling"

https://unsung.aresluna.org/traditional-but-puzzling/
2•colinprince•26m ago•0 comments

Show HN: Sill for Android – control your Mac's windows from an Android phone

https://github.com/rahimnathwani/sill-android
2•rahimnathwani•30m ago•0 comments

Gnome OS Preview

https://blogs.gnome.org/adrianvovk/2026/10/16/gnome-os-preview/
3•wirybeige•30m ago•0 comments

Fail Closed: Human-AI Research Without Pretending to Know

https://www.amazon.com/dp/B0HMJWK5L8
2•AlexanjaSenke•34m ago•0 comments

How the Claude Code team does code review

https://claude.dev/blog/solving-code-review-at-anthropic/
35•pretext•38m ago•29 comments

I Gave 100 Fly Brains Their Own Computers to Play Doom

https://namespace.so/blog/fly-brains-play-doom
2•nicolasd•38m ago•0 comments

Prisov – Private AI for Enterprise Workflows [video]

https://www.youtube.com/watch?v=9SyK9l8wi2A
2•harshalmore31•39m ago•0 comments