The best Hacker News stories from Show from the past week

from

All Show

from the past*

day week month

with these search terms:

Create follow Go back

Latest posts:

Show HN: Vibe coding a bookshelf with Claude Code

Show HN: Z80-μLM, a 'Conversational AI' That Fits in 40KB

How small can a language model be while still doing something useful? I wanted to find out, and had some spare time over the holidays.<p>Z80-μLM is a character-level language model with 2-bit quantized weights ({-2,-1,0,+1}) that runs on a Z80 with 64KB RAM. The entire thing: inference, weights, chat UI, it all fits in a 40KB .COM file that you can run in a CP/M emulator and hopefully even real hardware!<p>It won't write your emails, but it can be trained to play a stripped down version of 20 Questions, and is sometimes able to maintain the illusion of having simple but terse conversations with a distinct personality.<p>--<p>The extreme constraints nerd-sniped me and forced interesting trade-offs: trigram hashing (typo-tolerant, loses word order), 16-bit integer math, and some careful massaging of the training data meant I could keep the examples 'interesting'.<p>The key was quantization-aware training that accurately models the inference code limitations. The training loop runs both float and integer-quantized forward passes in parallel, scoring the model on how well its knowledge survives quantization. The weights are progressively pushed toward the 2-bit grid using straight-through estimators, with overflow penalties matching the Z80's 16-bit accumulator limits. By the end of training, the model has already adapted to its constraints, so no post-hoc quantization collapse.<p>Eventually I ended up spending a few dollars on Claude API to generate 20 questions data (see examples/guess/GUESS.COM), I hope Anthropic won't send me a C&D for distilling their model against the ToS ;P<p>But anyway, happy code-golf season everybody :)

Show HN: Z80-μLM, a 'Conversational AI' That Fits in 40KB

How small can a language model be while still doing something useful? I wanted to find out, and had some spare time over the holidays.<p>Z80-μLM is a character-level language model with 2-bit quantized weights ({-2,-1,0,+1}) that runs on a Z80 with 64KB RAM. The entire thing: inference, weights, chat UI, it all fits in a 40KB .COM file that you can run in a CP/M emulator and hopefully even real hardware!<p>It won't write your emails, but it can be trained to play a stripped down version of 20 Questions, and is sometimes able to maintain the illusion of having simple but terse conversations with a distinct personality.<p>--<p>The extreme constraints nerd-sniped me and forced interesting trade-offs: trigram hashing (typo-tolerant, loses word order), 16-bit integer math, and some careful massaging of the training data meant I could keep the examples 'interesting'.<p>The key was quantization-aware training that accurately models the inference code limitations. The training loop runs both float and integer-quantized forward passes in parallel, scoring the model on how well its knowledge survives quantization. The weights are progressively pushed toward the 2-bit grid using straight-through estimators, with overflow penalties matching the Z80's 16-bit accumulator limits. By the end of training, the model has already adapted to its constraints, so no post-hoc quantization collapse.<p>Eventually I ended up spending a few dollars on Claude API to generate 20 questions data (see examples/guess/GUESS.COM), I hope Anthropic won't send me a C&D for distilling their model against the ToS ;P<p>But anyway, happy code-golf season everybody :)

Show HN: Mysti – Claude, Codex, and Gemini debate your code, then synthesize

Hey HN! I'm Baha, creator of Mysti.<p>The problem: I pay for Claude Pro, ChatGPT Plus, and Gemini but only one could help at a time. On tricky architecture decisions, I wanted a second opinion.<p>The solution: Mysti lets you pick any two AI agents (Claude Code, Codex, Gemini) to collaborate. They each analyze your request, debate approaches, then synthesize the best solution.<p>Your prompt → Agent 1 analyzes → Agent 2 analyzes → Discussion → Synthesized solution<p>Why this matters: each model has different training and blind spots. Two perspectives catch edge cases one would miss. It's like pair programming with two senior devs who actually discuss before answering.<p>What you get: * Use your existing subscriptions (no new accounts, just your CLI tools) * 16 personas (Architect, Debugger, Security Expert, etc) * Full permission control from read-only to autonomous * Unified context when switching agents<p>Tech: TypeScript, VS Code Extension API, shells out to claude-code/codex-cli/gemini-cli<p>License: BSL 1.1, free for personal and educational use, converts to MIT in 2030 (would love input on this, does it make sense to just go MIT?)<p>GitHub: <a href="https://github.com/DeepMyst/Mysti" rel="nofollow">https://github.com/DeepMyst/Mysti</a><p>Would love feedback on the brainstorm mode. Is multi-agent collaboration actually useful or am I just solving my own niche problem?

Show HN: Ez FFmpeg – Video editing in plain English

I built a CLI tool that lets you do common video/audio operations without remembering ffmpeg syntax.<p>Instead of: ffmpeg -i video.mp4 -vf "fps=15,scale=480:-1:flags=lanczos" -loop 0 output.gif<p>You write: ff convert video.mp4 to gif<p>More examples: ff compress video.mp4 to 10mb ff trim video.mp4 from 0:30 to 1:00 ff extract audio from video.mp4 ff resize video.mp4 to 720p ff speed up video.mp4 by 2x ff reverse video.mp4<p>There are similar tools that use LLMs (wtffmpeg, llmpeg, ai-ffmpeg-cli), but they require API keys, cost money, and have latency.<p>Ez FFmpeg is different: - No AI – just regex pattern matching - Instant – no API calls - Free – no tokens - Offline – works without internet<p>It handles ~20 common operations that cover 90% of what developers actually do with ffmpeg. For edge cases, you still need ffmpeg directly.<p>Interactive mode (just type ff) shows media files in your current folder with typeahead search.<p>npm install -g ezff

Show HN: Ez FFmpeg – Video editing in plain English

I built a CLI tool that lets you do common video/audio operations without remembering ffmpeg syntax.<p>Instead of: ffmpeg -i video.mp4 -vf "fps=15,scale=480:-1:flags=lanczos" -loop 0 output.gif<p>You write: ff convert video.mp4 to gif<p>More examples: ff compress video.mp4 to 10mb ff trim video.mp4 from 0:30 to 1:00 ff extract audio from video.mp4 ff resize video.mp4 to 720p ff speed up video.mp4 by 2x ff reverse video.mp4<p>There are similar tools that use LLMs (wtffmpeg, llmpeg, ai-ffmpeg-cli), but they require API keys, cost money, and have latency.<p>Ez FFmpeg is different: - No AI – just regex pattern matching - Instant – no API calls - Free – no tokens - Offline – works without internet<p>It handles ~20 common operations that cover 90% of what developers actually do with ffmpeg. For edge cases, you still need ffmpeg directly.<p>Interactive mode (just type ff) shows media files in your current folder with typeahead search.<p>npm install -g ezff

Show HN: Witr – Explain why a process is running on your Linux system

Hi HN,<p>I built a small Linux CLI tool called witr (Why Is This Running?).<p>The idea came from a situation most of us have hit: you log into a machine, see a process or port running, and immediately wonder why it exists, who started it, and what is keeping it alive right now.<p>witr traces a process, service, or port back to its origin and responsibility chain and explains it in a way that’s quick to read, especially when you’re debugging under pressure.<p>This is v0.1.0. It’s intentionally small and focused. Feedback, criticism, and edge cases are very welcome.<p>Repo: <a href="https://github.com/pranshuparmar/witr" rel="nofollow">https://github.com/pranshuparmar/witr</a>

Show HN: Witr – Explain why a process is running on your Linux system

Hi HN,<p>I built a small Linux CLI tool called witr (Why Is This Running?).<p>The idea came from a situation most of us have hit: you log into a machine, see a process or port running, and immediately wonder why it exists, who started it, and what is keeping it alive right now.<p>witr traces a process, service, or port back to its origin and responsibility chain and explains it in a way that’s quick to read, especially when you’re debugging under pressure.<p>This is v0.1.0. It’s intentionally small and focused. Feedback, criticism, and edge cases are very welcome.<p>Repo: <a href="https://github.com/pranshuparmar/witr" rel="nofollow">https://github.com/pranshuparmar/witr</a>

Show HN: Gaming Couch – a local multiplayer party game platform for 8 players

Hi HN,<p>I’ve been working on Gaming Couch, a web-based game platform where up to 8 players use their smartphones as controllers to play real-time action mini-games on a central browser screen.<p>TL;DR:<p>- 18 competitive mini-games for up to 8 players<p>- Runs entirely in the browser<p>- Phones act as controllers (no apps, no accounts required)<p>- Focused on fast, chaotic, real-time party games (no trivia)<p>- Currently in public early access<p>Try it here: <a href="https://gamingcouch.com" rel="nofollow">https://gamingcouch.com</a>. Open the link on a computer, host a session, scan the QR code with your phone(s) and play!<p>What is it?<p>Gaming Couch is a party game platform where friends play short competitive action games together on one screen, using their phones as controllers (there's also support for physical gamepads if that's more your thing!)<p>I intentionally avoided trivia and text-heavy games. Many people don’t write or read English fluently, and I wanted games where reaction, timing, and chaos matter more than spelling.<p>It’s currently in early public access with 18 mini-games, all made by me and a two friends. All game rounds last ~1 minute, scores carry over, and after each round players vote on the next game. If you’re solo, 3 games support bots, but it’s best with a full couch of people as half the fun comes from the social aspect of playing together!<p>Why I built it:<p>For the last 15+ plus years, me and my friends have loved video game nights but organizing them has always been a PITA when you have more than 4 people playing:<p>- Different games were under different Steam accounts requiring downloads and installation.<p>- Extra controllers were missing (somebody forgot to bring theirs) or they wouldn’t pair.<p>- Consoles were expensive and not always available if we were on the road.<p>Once I started building it, other dev friends asked if they could make games for it too, which led me to realize this could also be a platform for small party games, especially for gamejam devs who don’t want to or have time to build multiplayer infrastructure from scratch. This is why supporting third-party games is the next major feature I’m working on.<p>Tech stack:<p>- Games run locally in the host’s browser (no streaming of games)<p>- Phones connect via WebRTC to the host session (1–10ms latency in ideal conditions with P2P connection)<p>- Fallback to TURN when direct P2P connection isn’t possible e.g. due to strict firewall settings in corporate networks or use of VPN's<p>- Website/Platform made with React + TypeScript<p>- Existing games made with Unity or just plain JS/TS.<p>- Backend: Supabase (Postgres + auth only, currently only used for optional user accounts)<p>How is it different from e.g. Jackbox, Airconsole or Nintendo?<p>Jackbox is absolutely great, but it’s heavily dependent on English literacy and "being funny" on the spot. I wanted something focused on fast, chaotic, real-time action games that work even if your friends speak different languages or just want to smash buttons. Also, I'm not a fan of their party pack model...<p>AirConsole is the most well known comparison to Gaming Couch in terms of technology and execution, but I feel there is a gap for a curated experience where the UI is unified, rounds are 60 seconds, and the competitive "meta-game" (scoreboards/voting) is baked into the platform. And in any case AirConsole was acquired by a car-software company and have pivoted their focus from couch gaming toward in-car entertainment.<p>Nintendo games are usually the gold standard in the party game category but the HW and games cost so much! With Gaming Couch, I want to keep the accessibility threshold as low as possible so everyone is able to play without upfront HW or SW costs.<p>What do you think of this? Are you an interested player or perhaps a developer who has had an idea to develop a fun 8 player mini-game but has been daunted by the idea thus far?

Show HN: Gaming Couch – a local multiplayer party game platform for 8 players

Hi HN,<p>I’ve been working on Gaming Couch, a web-based game platform where up to 8 players use their smartphones as controllers to play real-time action mini-games on a central browser screen.<p>TL;DR:<p>- 18 competitive mini-games for up to 8 players<p>- Runs entirely in the browser<p>- Phones act as controllers (no apps, no accounts required)<p>- Focused on fast, chaotic, real-time party games (no trivia)<p>- Currently in public early access<p>Try it here: <a href="https://gamingcouch.com" rel="nofollow">https://gamingcouch.com</a>. Open the link on a computer, host a session, scan the QR code with your phone(s) and play!<p>What is it?<p>Gaming Couch is a party game platform where friends play short competitive action games together on one screen, using their phones as controllers (there's also support for physical gamepads if that's more your thing!)<p>I intentionally avoided trivia and text-heavy games. Many people don’t write or read English fluently, and I wanted games where reaction, timing, and chaos matter more than spelling.<p>It’s currently in early public access with 18 mini-games, all made by me and a two friends. All game rounds last ~1 minute, scores carry over, and after each round players vote on the next game. If you’re solo, 3 games support bots, but it’s best with a full couch of people as half the fun comes from the social aspect of playing together!<p>Why I built it:<p>For the last 15+ plus years, me and my friends have loved video game nights but organizing them has always been a PITA when you have more than 4 people playing:<p>- Different games were under different Steam accounts requiring downloads and installation.<p>- Extra controllers were missing (somebody forgot to bring theirs) or they wouldn’t pair.<p>- Consoles were expensive and not always available if we were on the road.<p>Once I started building it, other dev friends asked if they could make games for it too, which led me to realize this could also be a platform for small party games, especially for gamejam devs who don’t want to or have time to build multiplayer infrastructure from scratch. This is why supporting third-party games is the next major feature I’m working on.<p>Tech stack:<p>- Games run locally in the host’s browser (no streaming of games)<p>- Phones connect via WebRTC to the host session (1–10ms latency in ideal conditions with P2P connection)<p>- Fallback to TURN when direct P2P connection isn’t possible e.g. due to strict firewall settings in corporate networks or use of VPN's<p>- Website/Platform made with React + TypeScript<p>- Existing games made with Unity or just plain JS/TS.<p>- Backend: Supabase (Postgres + auth only, currently only used for optional user accounts)<p>How is it different from e.g. Jackbox, Airconsole or Nintendo?<p>Jackbox is absolutely great, but it’s heavily dependent on English literacy and "being funny" on the spot. I wanted something focused on fast, chaotic, real-time action games that work even if your friends speak different languages or just want to smash buttons. Also, I'm not a fan of their party pack model...<p>AirConsole is the most well known comparison to Gaming Couch in terms of technology and execution, but I feel there is a gap for a curated experience where the UI is unified, rounds are 60 seconds, and the competitive "meta-game" (scoreboards/voting) is baked into the platform. And in any case AirConsole was acquired by a car-software company and have pivoted their focus from couch gaming toward in-car entertainment.<p>Nintendo games are usually the gold standard in the party game category but the HW and games cost so much! With Gaming Couch, I want to keep the accessibility threshold as low as possible so everyone is able to play without upfront HW or SW costs.<p>What do you think of this? Are you an interested player or perhaps a developer who has had an idea to develop a fun 8 player mini-game but has been daunted by the idea thus far?

Show HN: Vibium – Browser automation for AI and humans, by Selenium's creator

i started the selenium project 21 years ago. vibium is what i'd build if i started over today with ai agents in mind. go binary under the hood (handles browser, bidi, mcp) but devs never see it. just npm install vibium. python/java coming. for claude code: claude mcp add vibium -- npx -y vibium v1 ships today. ama.

Show HN: Vibium – Browser automation for AI and humans, by Selenium's creator

i started the selenium project 21 years ago. vibium is what i'd build if i started over today with ai agents in mind. go binary under the hood (handles browser, bidi, mcp) but devs never see it. just npm install vibium. python/java coming. for claude code: claude mcp add vibium -- npx -y vibium v1 ships today. ama.

Show HN: Minimalist editor that lives in browser, stores everything in the URL

I wanted to see how far I could go building a notes app using only what modern browsers already provide – no frameworks, no storage APIs, no build step.<p>What it does:<p>Single HTML file, no deps, 111 loc<p>Notes live in the URL hash (shareable links!)<p>Auto-compressed with CompressionStream<p>Plain-text editor (contenteditable)<p>History support<p>Page title from first # heading<p>Respects light/dark mode<p>No storage, cookies, or tracking<p>The entire app is the page source.<p><a href="https://textarea.my/" rel="nofollow">https://textarea.my/</a>

Show HN: Minimalist editor that lives in browser, stores everything in the URL

I wanted to see how far I could go building a notes app using only what modern browsers already provide – no frameworks, no storage APIs, no build step.<p>What it does:<p>Single HTML file, no deps, 111 loc<p>Notes live in the URL hash (shareable links!)<p>Auto-compressed with CompressionStream<p>Plain-text editor (contenteditable)<p>History support<p>Page title from first # heading<p>Respects light/dark mode<p>No storage, cookies, or tracking<p>The entire app is the page source.<p><a href="https://textarea.my/" rel="nofollow">https://textarea.my/</a>

Show HN: CineCLI – Browse and torrent movies directly from your terminal

Hi HN<p>I built CineCLI — a cross-platform terminal app to browse movies, view details, and open torrents directly in your system torrent client.<p>Features: - Search movies from the terminal - Rich UI with ratings, runtime, genres - Interactive & non-interactive modes - Magnet handling via system default client - Linux/macOS/Windows support - No ads, no tracking<p>GitHub: <a href="https://github.com/eyeblech/cinecli" rel="nofollow">https://github.com/eyeblech/cinecli</a> PyPI: <a href="https://pypi.org/project/cinecli/" rel="nofollow">https://pypi.org/project/cinecli/</a><p>Would love feedback from terminal + Python folks

Show HN: CineCLI – Browse and torrent movies directly from your terminal

Hi HN<p>I built CineCLI — a cross-platform terminal app to browse movies, view details, and open torrents directly in your system torrent client.<p>Features: - Search movies from the terminal - Rich UI with ratings, runtime, genres - Interactive & non-interactive modes - Magnet handling via system default client - Linux/macOS/Windows support - No ads, no tracking<p>GitHub: <a href="https://github.com/eyeblech/cinecli" rel="nofollow">https://github.com/eyeblech/cinecli</a> PyPI: <a href="https://pypi.org/project/cinecli/" rel="nofollow">https://pypi.org/project/cinecli/</a><p>Would love feedback from terminal + Python folks

Show HN: WalletWallet – create Apple passes from anything

I got my Apple developer certificate and built a simple app to solve a problem I had. One shop I buy from doesn't have Apple Wallet passes. Since you need signed certificates to build these very simple things, I created a minimal app that signs them. It's available if you need it too. It won't scan cards with AI - you manually enter the barcode, which I think makes it less prone to error.

Show HN: WalletWallet – create Apple passes from anything

I got my Apple developer certificate and built a simple app to solve a problem I had. One shop I buy from doesn't have Apple Wallet passes. Since you need signed certificates to build these very simple things, I created a minimal app that signs them. It's available if you need it too. It won't scan cards with AI - you manually enter the barcode, which I think makes it less prone to error.

Show HN: Books mentioned on Hacker News in 2025

Show HN: Books mentioned on Hacker News in 2025