The best Hacker News stories from Show from the past week
Latest posts:
Show HN: Encarta 95 Web
Encarta is super nostalgic, and I was surprised it wasn't available to use online. So I embarked on the difficult task of making it so.<p>The way this works is the most convoluted, nonsense and brittle system ever... but it works as a statically hosted gh-pages site!<p>Please note: All the material on this site is freely available through the internet archive (see github readme)<p>Github repo -> <a href="https://github.com/benwinding/encarta_95_web" rel="nofollow">https://github.com/benwinding/encarta_95_web</a>
Show HN: Hob - Open-source, on device alternative to Claude Cowork and ChatGPT
Hey HN, I'm Christian and this is a new OSS project I've been noodling on for the last couple of weeks. Hob is a free, MIT-licensed desktop app for Mac, Windows and Linux that works on your files the way the Claude and ChatGPT desktop apps do, except the model runs on your own computer. It uses open-weight models (currently Gemma 4 and gpt-oss). All your chats and files never leave your system. The only things that go online are web search (can be turned off in settings) and any MCP connectors you add yourself.<p>Main features:<p>- Automatically finds the best open-weight model for your hardware on first install (Mac/Linux/Windows)<p>- Writes sandboxed Python code for ad hoc tools (data processing, document conversion, graphs etc.)<p>- Built-in web search (powered by SearXNG)<p>- MCP support<p>- Image recognition<p>- CRUD to your local filesystem<p>- Works fully offline with no internet<p>Before building this I tried every local AI app I could find, but it seemed each one had different problems. The main thing I noticed is that none of them could do what the ChatGPT and Claude apps do with your files, e.g. read and change files in a folder, look at images, run code in a sandbox, draw a chart. The most advanced thing any of them did was paste a plain text file into a small context window, which basically failed as soon as you had a file larger than a few KB. I couldn't find anything that could easily parse images and bigger PDFs.<p>The OSS solutions that were more advanced all seemed to default to hosted models (either OSS or closed), which defeats the entire point of the privacy that on-device OSS offers. They also seemed to push towards an account/open-core style business.<p>I build it using Electron bridged with llama.cpp in its own process. On initial load, the system will try and figure out if you have a compatible graphics card and if not default to CPU-only (I ran into a lot of issues of graphics card beeing seemingly compatible but then crashing on first query, there's a whole saga to this lol). Code execution happens in a sandbox that can only write to the chat's folder and has no network unless asked (sandbox-exec on macOS, Landlock on Linux and a low-integrity process on Windows). Images go to a small Qwen3-VL model. Dictation is a local Whisper model.<p>The hardest part was definitely to get the smaller models to use tools reliably. Ran into a ton of issues of 4B models happily saying "here's your chart" without ever calling the chart tool or get stuck in an infinite call loop. I spent a fair amount of the work building guards vs tweaking prompts (that didn't really work for small models).<p>I understand the pushback that local models are weaker than frontier cloud models. This is true, but for most tasks for most normal people, I'd argue you rarely need the newest frontier model. From my own evals, systems with 16+ GB of RAM do really well, and if you have a GPU or more RAM, you get very close to what you'll get out of the box with a frontier hosted model.<p>Excited to hear what you think and see how it runs on other systems!
Show HN: Hob - Open-source, on device alternative to Claude Cowork and ChatGPT
Hey HN, I'm Christian and this is a new OSS project I've been noodling on for the last couple of weeks. Hob is a free, MIT-licensed desktop app for Mac, Windows and Linux that works on your files the way the Claude and ChatGPT desktop apps do, except the model runs on your own computer. It uses open-weight models (currently Gemma 4 and gpt-oss). All your chats and files never leave your system. The only things that go online are web search (can be turned off in settings) and any MCP connectors you add yourself.<p>Main features:<p>- Automatically finds the best open-weight model for your hardware on first install (Mac/Linux/Windows)<p>- Writes sandboxed Python code for ad hoc tools (data processing, document conversion, graphs etc.)<p>- Built-in web search (powered by SearXNG)<p>- MCP support<p>- Image recognition<p>- CRUD to your local filesystem<p>- Works fully offline with no internet<p>Before building this I tried every local AI app I could find, but it seemed each one had different problems. The main thing I noticed is that none of them could do what the ChatGPT and Claude apps do with your files, e.g. read and change files in a folder, look at images, run code in a sandbox, draw a chart. The most advanced thing any of them did was paste a plain text file into a small context window, which basically failed as soon as you had a file larger than a few KB. I couldn't find anything that could easily parse images and bigger PDFs.<p>The OSS solutions that were more advanced all seemed to default to hosted models (either OSS or closed), which defeats the entire point of the privacy that on-device OSS offers. They also seemed to push towards an account/open-core style business.<p>I build it using Electron bridged with llama.cpp in its own process. On initial load, the system will try and figure out if you have a compatible graphics card and if not default to CPU-only (I ran into a lot of issues of graphics card beeing seemingly compatible but then crashing on first query, there's a whole saga to this lol). Code execution happens in a sandbox that can only write to the chat's folder and has no network unless asked (sandbox-exec on macOS, Landlock on Linux and a low-integrity process on Windows). Images go to a small Qwen3-VL model. Dictation is a local Whisper model.<p>The hardest part was definitely to get the smaller models to use tools reliably. Ran into a ton of issues of 4B models happily saying "here's your chart" without ever calling the chart tool or get stuck in an infinite call loop. I spent a fair amount of the work building guards vs tweaking prompts (that didn't really work for small models).<p>I understand the pushback that local models are weaker than frontier cloud models. This is true, but for most tasks for most normal people, I'd argue you rarely need the newest frontier model. From my own evals, systems with 16+ GB of RAM do really well, and if you have a GPU or more RAM, you get very close to what you'll get out of the box with a frontier hosted model.<p>Excited to hear what you think and see how it runs on other systems!
Show HN: A Truly Modular Keyboard
Show HN: A Truly Modular Keyboard
Show HN: TitleArt – Make text like it's 1997
Show HN: TitleArt – Make text like it's 1997
Show HN: I put employees inside S&P 500 market-cap bubbles
I was playing around with visualising the S&P 500 by market cap and decided to add employees to each company's bubble.<p>What surprised me was how empty some of the biggest companies looked.<p>Nvidia is worth $5.2 trillion but employs only 42,000 people. That's around $124 million in market cap per employee. Meanwhile, Accenture employs 814,000 people and is worth $146 billion.<p>Each stick figure represents 10,000 employees. You can zoom in and explore individual companies.<p>I wrote a bit more about what I found here: <a href="https://bigwhalehunting.com/blog/workforce/" rel="nofollow">https://bigwhalehunting.com/blog/workforce/</a><p>Would be interested to hear what you think.
Show HN: I put employees inside S&P 500 market-cap bubbles
I was playing around with visualising the S&P 500 by market cap and decided to add employees to each company's bubble.<p>What surprised me was how empty some of the biggest companies looked.<p>Nvidia is worth $5.2 trillion but employs only 42,000 people. That's around $124 million in market cap per employee. Meanwhile, Accenture employs 814,000 people and is worth $146 billion.<p>Each stick figure represents 10,000 employees. You can zoom in and explore individual companies.<p>I wrote a bit more about what I found here: <a href="https://bigwhalehunting.com/blog/workforce/" rel="nofollow">https://bigwhalehunting.com/blog/workforce/</a><p>Would be interested to hear what you think.
Show HN: Steal My Bitcoin
Made this because I'm curious if anyone out there has figured out how to actually use this PHP RCE we spent the last few days panicking over. Have fun!
Show HN: Steal My Bitcoin
Made this because I'm curious if anyone out there has figured out how to actually use this PHP RCE we spent the last few days panicking over. Have fun!
Show HN: OnKey – I'm a bad singer, so I built an app to show where I'm off-key
Hi all<p>I was pretty terrible at singing, now I’d say I’m bad but not terrible. This is because I’ve been using a singing app (won’t name it here) that scores me on my pitch and timing throughout the song so that I can improve.<p>Lately however, the app I’ve been using has flooded their song list with songs that aren’t originals. I didn’t like this at all as for me, it’s easier to try to improve when you’re singing along to the original.<p>So, I asked an agent (Claude Code) if it could write some code to listen to Spotify and also listen to me and see if it could detect both pitches and score me. It impressed me by writing some Python and showing me my pitch in the terminal!<p>It couldn’t go all the way though and detect the singer’s pitch in real time.<p>So I found the next best setup and built OnKey, a Mac app that needs to listen to the song once (it doesn’t save it) and then when it’s learnt the song (takes a couple of minutes), you can sing along to the pitch track that it’s learnt. OnKey is free to use and open source under the MIT license.<p>It can listen to Spotify, Apple Music, and YouTube playing in your browser with the right permissions.<p>During the build I found that someone had done something similar in Nightingale, but this requires a library of songs exported to disk and who has this in 2026 (maybe some of you ;) - but we need to help the normies too).<p>You can check out a demo video on the site of me singing (warning given) here: <a href="https://singonkey.app" rel="nofollow">https://singonkey.app</a><p>The app sends out to find the song info (short Shazam fingerprint, and looks up song details), lyrics (from LRCLIB), and checks for updates.<p>The app never sends the music (it doesn’t save this to disk) and your voice. Everything you’d expect to stay private does.<p>It’s still weak in a few places:<p>- when the song’s backing is noisy it’s hard to pick the pitch track out<p>- sometimes it gets a note an octave off<p>- it’s limited to Apple Silicon Macs, M1 or later, and only macOS 14.2 up because extra drivers aren’t required past this point<p>- the scoring is built for beginners like me so it’s fairly easy on you<p>Would love for people to try it out and let me know what works and what doesn’t for them!
Run full iOS 16.5 firmware on Android Devices
<a href="https://hackios-web.vercel.app/en" rel="nofollow">https://hackios-web.vercel.app/en</a>
Show HN: SimCity, but with Permits
Show HN: Qwen3-4B with the AI writing tics trained out
Show HN: WhatsApp MCP running free on Cloudflare
I wanted my humble Claude on mobile to access my WhatsApp messages but the only options I could find either required WhatsApp Business or an actual server to run Baileys on.<p>I wanted to run it for free on ClaudeFlare and with my personal WhatsApp account, so I built this.<p>Important feature - it connects to your Google account as well as WhatsApp to grab the names of all your WhatsApp contacts, otherwise you get only phone numbers for many of them and search doesn't work properly.<p>This is a single user install because it's both gray area and very private - so point your agent at the repo and let it set things up.
Show HN: The Zelda timeline to catch up before the Ocarina of Time remake
I made this ahead of the Ocarina of Time remake which comes out in 17 days. I remember the feeling of playing it as a kid but I hadn’t kept up with the story so I built a timeline to catch up before the Door of Time reopens.<p>Nintendo’s own timeline is the authority here (Hyrule Historia and their chronicle page). The fun part was rethinking how to visualize it. It branches at the end of Ocarina into three eras depending on whether Link wins or loses, so each era is its own thread on the cloth. The threads meet again much later with Breath of the Wild.<p>I had a lot of help from Claude. I’d often thought about building things like this but never found the time between the day job and the rest of life.
Show HN: Sphinx – A hook that quizzes you on your AI's code before shipping
FOSS, Basically it plugs into your existing coding agent workflow and every time you push to prod it'll throw up a questionnaire that checks if you understand how the code you're pushing works.
Show HN: Sphinx – A hook that quizzes you on your AI's code before shipping
FOSS, Basically it plugs into your existing coding agent workflow and every time you push to prod it'll throw up a questionnaire that checks if you understand how the code you're pushing works.
Show HN: High-quality duck audio synthesizer