frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Learning to code, or building side projects with AI help, this one's for you

https://codeslick.dev/learn
1•vitorlourenco•36s ago•0 comments

Effulgence RPG Engine [video]

https://www.youtube.com/watch?v=xFQOUe9S7dU
1•msuniverse2026•2m ago•0 comments

Five disciplines discovered the same math independently – none of them knew

https://freethemath.org
1•energyscholar•2m ago•1 comments

We Scanned an AI Assistant for Security Issues: 12,465 Vulnerabilities

https://codeslick.dev/blog/openclaw-security-audit
1•vitorlourenco•3m ago•0 comments

Amazon no longer defend cloud customers against video patent infringement claims

https://ipfray.com/amazon-no-longer-defends-cloud-customers-against-video-patent-infringement-cla...
1•ffworld•3m ago•0 comments

Show HN: Medinilla – an OCPP compliant .NET back end (partially done)

https://github.com/eliodecolli/Medinilla
2•rhcm•6m ago•0 comments

How Does AI Distribute the Pie? Large Language Models and the Ultimatum Game

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6157066
1•dkga•7m ago•1 comments

Resistance Infrastructure

https://www.profgalloway.com/resistance-infrastructure/
2•samizdis•11m ago•0 comments

Fire-juggling unicyclist caught performing on crossing

https://news.sky.com/story/fire-juggling-unicyclist-caught-performing-on-crossing-13504459
1•austinallegro•12m ago•0 comments

Restoring a lost 1981 Unix roguelike (protoHack) and preserving Hack 1.0.3

https://github.com/Critlist/protoHack
2•Critlist•13m ago•0 comments

GPS and Time Dilation – Special and General Relativity

https://philosophersview.com/gps-and-time-dilation/
1•mistyvales•17m ago•0 comments

Show HN: Witnessd – Prove human authorship via hardware-bound jitter seals

https://github.com/writerslogic/witnessd
1•davidcondrey•17m ago•1 comments

Show HN: I built a clawdbot that texts like your crush

https://14.israelfirew.co
2•IsruAlpha•19m ago•2 comments

Scientists reverse Alzheimer's in mice and restore memory (2025)

https://www.sciencedaily.com/releases/2025/12/251224032354.htm
1•walterbell•22m ago•0 comments

Compiling Prolog to Forth [pdf]

https://vfxforth.com/flag/jfar/vol4/no4/article4.pdf
1•todsacerdoti•23m ago•0 comments

Show HN: Cymatica – an experimental, meditative audiovisual app

https://apps.apple.com/us/app/cymatica-sounds-visualizer/id6748863721
1•_august•25m ago•0 comments

GitBlack: Tracing America's Foundation

https://gitblack.vercel.app/
2•martialg•25m ago•0 comments

Horizon-LM: A RAM-Centric Architecture for LLM Training

https://arxiv.org/abs/2602.04816
1•chrsw•25m ago•0 comments

We just ordered shawarma and fries from Cursor [video]

https://www.youtube.com/shorts/WALQOiugbWc
1•jeffreyjin•26m ago•1 comments

Correctio

https://rhetoric.byu.edu/Figures/C/correctio.htm
1•grantpitt•26m ago•0 comments

Trying to make an Automated Ecologist: A first pass through the Biotime dataset

https://chillphysicsenjoyer.substack.com/p/trying-to-make-an-automated-ecologist
1•crescit_eundo•30m ago•0 comments

Watch Ukraine's Minigun-Firing, Drone-Hunting Turboprop in Action

https://www.twz.com/air/watch-ukraines-minigun-firing-drone-hunting-turboprop-in-action
1•breve•31m ago•0 comments

Free Trial: AI Interviewer

https://ai-interviewer.nuvoice.ai/
1•sijain2•31m ago•0 comments

FDA intends to take action against non-FDA-approved GLP-1 drugs

https://www.fda.gov/news-events/press-announcements/fda-intends-take-action-against-non-fda-appro...
23•randycupertino•32m ago•14 comments

Supernote e-ink devices for writing like paper

https://supernote.eu/choose-your-product/
3•janandonly•35m ago•0 comments

We are QA Engineers now

https://serce.me/posts/2026-02-05-we-are-qa-engineers-now
1•SerCe•35m ago•0 comments

Show HN: Measuring how AI agent teams improve issue resolution on SWE-Verified

https://arxiv.org/abs/2602.01465
2•NBenkovich•35m ago•0 comments

Adversarial Reasoning: Multiagent World Models for Closing the Simulation Gap

https://www.latent.space/p/adversarial-reasoning
1•swyx•36m ago•0 comments

Show HN: Poddley.com – Follow people, not podcasts

https://poddley.com/guests/ana-kasparian/episodes
1•onesandofgrain•44m ago•0 comments

Layoffs Surge 118% in January – The Highest Since 2009

https://www.cnbc.com/2026/02/05/layoff-and-hiring-announcements-hit-their-worst-january-levels-si...
14•karakoram•44m ago•0 comments
Open in hackernews

Ask HN: Have you used Claude Code? Is it any good?

8•mbm•9mo ago
If you do use it, how does it fit into your workflow?

Comments

viraptor•9mo ago
Much worse than Cursor with Claude models in my experience. I'm getting many useless changes and things being reimplemented from scratch instead of moving files. Not impressed at all.
mbm•9mo ago
Interesting, thanks for sharing.
kasey_junk•9mo ago
Fwiw I’ve gotten really impressive results from Claude code and it’s the first time I’ve seen that with any agentic flow (including Cursor).

Sorry to give opposite anecdotes. It’s one of the things I find most irritating about AI right now.

muzani•9mo ago
Anecdotes are better than data, especially when that data is exclusively benchmarked on Python and code competitions instead of actual work.
disqard•9mo ago
I'm finding it to be a force-multiplier in my side-project work.

It needs careful oversight, for if you're too generous with it, it'll happily add tons of code into your codebase that will make it horrendously difficult to understand and debug later. As capable as it is, I find it prudent to keep it on a short leash.

tombot•9mo ago
It gets a lot better the more you adopt best practices, my first step is often just to ask it make a plan I can review before it touches anything.

https://www.anthropic.com/engineering/claude-code-best-pract...

nickisnoble•9mo ago
It's still not great, but it's better than anything else.

Best results when:

1. run /init and let it maintain a CLAUDE.md

2. Ask it to run checks + tests before / after every task, and add those commands to the "no permission needed list" – this improves quality by a lot

3. Ask it to do TDD, but manually check the actual test is correct

4. Every time it finishes something solid: git commit manually, /compact context (saves hella $$$ + seems to improve focus)

Honestly I treat it like a junior programmer I'm pairing with. If you pay attention, you can catch it being stupid early and get it back on track. Best when you know exactly what you want, it's just boring work. It's really good with clear instructions, eg "Refactor X -> Y, using {design pattern}."

mike_hearn•9mo ago
Yep. I've given it tools (scripts, not MCP) that give it access to my bug tracker. It's able to handle requests like "please fix bug CO-1234" and off it goes. I have another script that uses Gemini to create map files, and that helps Claude find the right parts of the codebase quickly. It's able to do quite sophisticated bug fixes in a mature and unusual codebase (it's a deployment tool for desktop apps), and the quality of the produced code is high.

One thing I miss from Aider is speech recognition integration. When you're alone at home it's great to be able to speak what you want instead of typing it.

I'm pretty happy with it. The next big upgrade would be deep IDE integration, complete with the ability to search the IDE indexes, navigate around using cross-references and the like.

LeafItAlone•9mo ago
Are you able to share you Gemini script? That sounds useful.