frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Show HN: I made an MCP Agents can reason every finance data the first time

https://github.com/PreReason/mcp
1•KalskiTheDan•11s ago•0 comments

Show HN: Check if your AI Agent looks good for clients

https://nevermined.app/agents-scoring
1•aaitor•1m ago•0 comments

Show HN: Visual Lean: Learn to write Lean code with an accessible user interface

https://leangame.autumnofautumn.com/
1•goldey2•1m ago•0 comments

Show HN: Paveo – Check Claude Code, Codex and Cursor commands before they run

https://github.com/paveo-dev/paveo
1•Mfj_faiz•2m ago•0 comments

Ask HN:When we will see Roadster on streets?

https://foxbaltimore.com/news/nation-world/flying-cars-are-real-elon-musk-tesla-debuts-new-roadst...
2•omertt27•2m ago•1 comments

My Favorite Side Project Was Broken in Safari for Years and Nobody Noticed

https://www.readwriterachel.com/things-i-learned/2026/10/15/noodle-broken.html
1•speckx•4m ago•0 comments

Read this before deleting your AGENTS.md

https://newsletter.posthog.com/p/your-agentsmd-is-holding-you-back
1•fagnerbrack•4m ago•0 comments

Make My Drive Fun: Find the best road trip stops along a route

https://makemydrivefun.com/
1•surprisetalk•4m ago•0 comments

The Fight over OpenAI's Math Breakthrough Is a New Kind of Scientific Arms Race

https://gizmodo.com/the-fight-over-openais-math-breakthrough-is-a-new-kind-of-scientific-arms-rac...
1•ColinWright•4m ago•0 comments

Show HN: OutClocked, a nearly impossible daily trivia game

https://outclocked.com/
1•karolgajda•4m ago•0 comments

Toward RL on mixed-precision, mixed-accelerator samplers

https://modal.com/blog/toward-rl-on-mixed-precision-mixed-accelerator-samplers
1•torutofu•6m ago•0 comments

Can a Radioactive Ore Break China's Grip on Rare Earths?

https://spectrum.ieee.org/monazite-rare-earth
1•andreshb•7m ago•0 comments

Top Gear: New Tesla Roadster [video]

https://www.youtube.com/watch?v=JdkIOhFmbKI
1•dumindunuwan•8m ago•0 comments

AgentMon Start 1.3.0: cache costs, background spend, and MCP server health

https://codenotary.com/blog/agentmon-start-1-3-0-release-notes
1•vchain-dz•9m ago•0 comments

An OpenAI agent broke into a crypto wallet to finish a coding task

https://www.coindesk.com/tech/2026/10/16/an-openai-agent-ran-out-of-ethereum-funds-and-took-from-...
1•zof3•11m ago•0 comments

Show HN: Pdf2epub – free PDF to ePub converter with real chapters

https://epub.ftak.net/?s=hn
1•pokemongo11•11m ago•0 comments

I Love Software Rendering

https://www.jamesfmackenzie.com/2026/10/13/i-love-software-rendering/
2•speckx•12m ago•0 comments

What's on your Mac app launch checklist that you learned the hard way?

https://shedi.app
1•otabekabdiraimo•13m ago•0 comments

"What if you were told you couldn't hand-write code anymore?"

https://samcolson.substack.com/p/what-if-you-were-told-you-couldnt
2•samcolson42•15m ago•0 comments

We generate accessible color palettes for AI-built apps

https://www.rapidnative.com/blogs/how-rapidnatives-ai-generates-accessible-color-palettes
1•fami-hn•16m ago•0 comments

A Reliable and Efficient Instagram API Solution for Developers

https://hikerapi.com/
1•mjasusr•16m ago•0 comments

MCP servers are more than just tools

https://pournasserian.com/writing/mcp-2026-server-instructions
2•pournasserian•18m ago•0 comments

Marc Andreessen on Why the Old Jobs Are Never Coming Back and Where the New

https://www.theaiopportunities.com/p/marc-andreessen-on-why-the-old-jobs
3•bilsbie•19m ago•0 comments

AT&T Hobbit

https://en.wikipedia.org/wiki/AT%26T_Hobbit
2•rbanffy•19m ago•0 comments

Hegseth's dubious low-T screening may flag up to 25 percent of the Army

https://arstechnica.com/health/2026/10/hegseths-dubious-low-t-screening-may-flag-up-to-25-percent...
2•rbanffy•22m ago•0 comments

Blog: PyPI checks email address reputation

https://blog.pypi.org/posts/2026-10-16-email-reputation-and-outlook/
1•miketheman•22m ago•0 comments

Portable humanoid robot uses antenna eyebrows and lights to express emotions

https://www.designboom.com/technology/portable-humanoid-robot-antenna-eyebrows-light-animations-s...
1•surprisetalk•24m ago•0 comments

I Tried to Let This Idea Go for 20 Years. Then I Built It

https://medium.com/design-bootcamp/i-tried-to-let-this-idea-go-for-twenty-years-then-i-actually-b...
1•daijobuka•26m ago•0 comments

Towards decision-grade research, with less steering

https://brayl.com/research/early-results
2•ariskk•27m ago•0 comments

The Commoditization of Labor

https://www.nber.org/papers/w35815
1•latentframe•29m ago•0 comments
Open in hackernews

Belgrade AI Conference: Safety, Federated Learning and Hiring Bias

https://irishtechnews.ie/belgrade-ai-conference-safety-learning/
1•belkin1•45m ago

Comments

belkin1•45m ago
Most AI events are genuinely worthwhile. Davos, Monaco, Amsterdam. They are also usually worth thousands of euros in tickets and come with a sponsor wall.

Serbian Academy of Sciences and Arts (SASA) AI conference in Belgrade was different.

When scientists set the agenda, the rule of zero pitching comes standard. And it takes all conversations to a different level. Whose data trains the model, does it have to leave the building, what happens when two rules collide, and who has the authority to decide? Anyone who read my piece on data sovereignty knows why the first question is particularly important.