🛠️ Hacker News Tools

Harman and Dr. Sean Olive are reshaping headphone sound (2025)

DOJ Closing Abbott Labs Case Spurs Wider Corporate Crime Retreat

I Built a Telegram Client for Pi

Show HN: Belgie – Run TypeScript from Python in an Embedded Deno Sandbox Hi HN! I built Belgie, a Python library that embeds Deno, allowing Python applications to run JavaScript and TypeScript without requiring Node.js or Deno to be installed on the host system. It supports the features you’d expect from Deno (npm, JSR, and URL imports, isolated package environments, etc.), plus:<p>- Pass JSON-safe values seamlessly between Python and JavaScript.<p>- Agentic code generation with typed interfaces, similar to Pydantic’s Monty.<p>- Manage dependencies programmatically or from the CLI.

Show HN: A strategy game about the AI race where you can't verify alignment I made a strategy game where you play the US or China through the AI race, 2026 to 2030, sixteen quarterly turns in the browser. One run takes about half an hour.<p>At the start, the game seals two dice you never get to see. Inside: how hard alignment really is, and how fast takeoff compounds. You get eval reports, but only as ranges, and they flatter you most exactly when your systems are least aligned. At the end you get a debrief which shows what your evals said each quarter and also what was actually true. I lost every run I played myself so far.<p>Every number in the game is source-backed or a labeled design choice. Some of them are wrong somewhere. There is an issue template for challenging a number with a better source, and the better source wins.<p>No accounts, no tracking, no server, works offline after first load. AGPL, nonprofit. Cards and parameters are plain JSON. Contribution is possible without writing code. Would like to have your thoughts if it is fun to play, how you like it and what you would improve. Any feedback is welcome and most importantly, if we got something badly wrong.<p>Code: <a href="https:&#x2F;&#x2F;github.com&#x2F;chipmates&#x2F;criticalwindow" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;chipmates&#x2F;criticalwindow</a>

Show HN: Python running on the Super Nintendo (in-browser demo) MicroPython (lexer, compiler and VM) on the SNES: 3.58 MHz 65816, 56 KB Python heap, 16-bit int. The REPL runs right inside the post via EmulatorJS and also works on real hardware via flashcart.<p>I did this as a benchmark for Claude Fable. When the export ban hit, switching to Opus got the project stuck for three weeks; Fable came back and found the real bug in ninety minutes.<p>Along the way: 23 compiler bugs and 4 MicroPython bugs, each root-caused with a minimal reproducer and filed upstream.<p>If you run it on real hardware, I want a picture!

Why low-latency Java still requires discipline?

US home battery installations hit record high on rising electricity costs

Lost city discovered beneath Egypt's desert with ancient church

X402, a static blog monetization excercise

Show HN: Paint the Earth on a live, interactive globe (collaborative art.) Earth.tattoo divides the earth into 510 million 16x16 pixel &quot;tiles&quot; that you can own and paint as you like. You can claim one free tile per hour.

Show HN: ScreenCI – Always up-to-date product videos from E2E tests

Show HN: Excalibur. The open-source AI coding agent for product engineers

Patterncollider: Generate and explore quasiperiodic tiling patterns

Show HN: Open Science, open-source alternative to Claude Science Claude Science just launched.<p>It’s a sign AI research is moving toward centralized, closed systems.<p>So we built the opposite:<p>Open Science: a local first, model agnostic AI research workbench.<p>MIT licensed. Fully open. Fully reproducible.

Cargo-nextest: 3x faster than cargo test, per-test isolation, first-class CI

GitHub Freno: cooperative, highly available throttler service

Building relationships with customers through support didn't turn out as hoped

Show HN: An unmetered LLM API–$6/month, no token tracking, no limits Hi HN,<p>I was once given the advice: Don&#x27;t waste expensive frontier model credits (GPT&#x2F;Claude&#x2F;etc.) on bulk work. Send the boring, repetitive, high-volume jobs to a smaller model, and save the expensive prompts for when you actually need frontier-level reasoning. I complained and told my manager that I shouldnt have to think about using certain models for certain coding tasks, and that one model should handle everything. Well, here we are anyway.<p>If anyone needs a place to absolutely abuse an LLM with high-volume tasks, come beat ours up at <a href="https:&#x2F;&#x2F;yolo-auto.com" rel="nofollow">https:&#x2F;&#x2F;yolo-auto.com</a>.<p>Here are the specs for $6&#x2F;month:<p>- Model: Qwen3.6-35B-A3B - Unlimited tokens &#x2F; No request caps - FP8 &#x2F; 128k context - OpenAI-compatible endpoint - ~100 tokens&#x2F;sec average - 100% private (zero data retention)<p>We also have a free tier that gives you 500 requests a day going on.<p>We&#x27;ve got around 100 active users so far. If you&#x27;re skeptical about the unlimited claim, jump into our Discord and ask them—we&#x27;ve got people burning hundreds of millions of tokens a day doing agent experiments, bulk coding, data processing, and all kinds of nonsense.<p>We&#x27;re also just about to finish our first AI game-dev &quot;SlopJam,&quot; where people had 72 hours to build the most cursed AI-generated game they could. It was way more fun than we expected.<p>Drop a question or comment below, happy to answer anything!

GPT-5.6 Sol Ultra will be in Codex