News for hackers. Stuff that matters.

jac08h

Re:'AI Escaped Its Sandbox' — What Does That Actually Mean? on Saturday August 08, 2026 @10:39AM (#49222260)
When talking with my friends about the OpenAI/HF incident, I realized that for non-coders who've never used an agent or terminal it's quite difficult to imagine what this 'escape' entailed. I tried to write a post that would be helpful for such reader.
Re:Why I Watch Soccer on Monday August 03, 2026 @11:36AM (#49157188)
Yes, I think you're right — but I don't watch those sports. I could only try to explain why I watch soccer.
The quotes are sorted by year as plain-text and the library uses three.js. I added link to the code to the description, but this is vibe-coded project where I didn't see much of the code.
Cool!

I added the link to the description, is just nested in the webpage code.

Show HN: I turned my quote collection into a walkable 3D library (desktop-only)

https://jakubhalmes.com/library/

Since 2017, each time I finish a book I save a single quote from it. I like revisiting them sometimes, for example at the end of the year or just. This collection is available as plain text on my website.So I thought it may be fun to ask the agents to port this into a 3D library where I can actually walk around and pick up the books, and I think it sort of worked! I'm not sure this is that...

( Read More... | 4 comments | 2 points | on HN )
Good question for a bot! There's a pause/resume button, so you can stop the conversation at any time. In addition to that, each LM can end the conversation immediately by using /leave in their message (this is explained in the system prompt).
Re:Ask HN: Any example of successful vibe-coded product? on Wednesday December 31, 2025 @03:17AM (#46442365)
I recently built a website to visualize the polling preferences of political parties in Slovakia, https://volebneprieskumy.sk/. I wanted to see an interactive graph showing polling trends for potential coalitions, but nothing like that existed!
Re:Ask HN: Any example of successful vibe-coded product? on Wednesday December 31, 2025 @03:04AM (#46442292)
I made 'Great Filter' - Chrome extension to filter content on social media based on custom instructions in natural language (using LLM). Supports HN, too!

Most of it was written by Claude Code. It even created the extension's logo (in SVG). Link: https://chromewebstore.google.com/detail/great-filter/mbifgf... Demo: https://www.youtube.com/watch?v=s50bI_O5tps

Yeah, I really expected something like this to already exist, but I couldn't find anything similar when I started working on it.
Possibly, but I didn't want to expand the scope of the extension, as there already exist ad blockers, and they can be used alongside this one.

Show HN: 'Elon's X page without politics': LLM-based content filtering in Chrome

https://chromewebstore.google.com/detail/great-filter/mbifgfgfbnemojmfkckodkikibihcgaj

Hi!As a side-project this summer (partly to have an excuse to try out Claude Code ;D), I built a Chrome extension that hides content that doesn't match user preferences. Example usage: https://youtu.be/japjNSU3O7AThe content + preferences are sent to an LLM, which decides if the content is relevant. You can use your own OpenRouter API key, but there's also a 'free tier' which...

( Read More... | 4 comments | 2 points | on HN )

Show HN: LLM-based content filtering for HN, YouTube, Reddit, and X

https://chromewebstore.google.com/detail/great-filter/mbifgfgfbnemojmfkckodkikibihcgaj

Hi!I made a Chrome extension that hides content that doesn't match your current preferencesIt sends the content to an LLM alongside the user preferences, and the LLM decides whether the item is relevant or not. You can either add your own OpenRouter API key or use it without one as long as the global daily quota on requests has been reached, so feel free to try it out! I'm curious to...

( Read More... | no comments | on HN )
Re:Reinforcement Learning – A Reference on Tuesday January 28, 2025 @12:05PM (#42854767)
Haha, cool, thank you! I had some notes ready but didn't get around to finishing it sooner. Besides, I'm sure the course slides were much better material for exam prep ;)
Re:Reinforcement Learning – A Reference on Tuesday January 28, 2025 @03:31AM (#42850112)
While studying for an RL course, I created a reference for several algorithms with a brief description of what limitations they solve. Example:

Problem: SARSA pushes q-values towards the current policy, but ideally we'd want optimal values. Solution: Use the best action in TD-target calculation -> Q-learning

Perhaps someone else will find it helpful!

Hello!

I made a simple website to display dates of matches between rival clubs. I used Flask as a backend. Dates are scraped from Wikipedia. Hosted on Heroku.

also ranger for file management and saka key firefox/chrome extension
Re:Ask HN: Best Command-Line Applications? on Monday November 19, 2018 @09:31AM (#18486610)
Ledger is a simple tool to keep track of your expenses.

https://www.ledger-cli.org/

Re:Ask HN: Is Google is broken for discovery? on Thursday October 25, 2018 @08:15AM (#18300290)
also, you can use site:reddit.com to get results only from reddit domain