Reading up on open-source
100 deep · digging since nov 19, 25
- Just say no to OpenAI and Anthropic
The author urges rejecting OpenAI and Anthropic due to privacy risks and oligarchic control, recommending open-weight LLMs served via private enclave providers like Tinfoil or Maple.
- Vibe-Coded Browser Ports Of Games Work Perfectly, Unfortunately
Claude Opus 5.5 enables rapid AI decompilation of games, producing functional browser ports of titles like Call of Duty: Black Ops and Halo, raising IP concerns.
- RepRap: Blog: RepRapMicron Success! - A decent 500μm (1/2mm) Benchy
The author successfully printed a 500‑micrometer‑tall Benchy using the RepRapMicron PIKA V0.02 printer and eSun Standard Resin in under three hours.
- Show HN: Quake ported to safe Rust, playable in browser
A developer has ported the classic Quake engine to safe Rust, compiling it to WebAssembly so the game runs directly in a browser.
- Supabase is acquiring Turso
Supabase announced it is acquiring Turso, the edge‑hosted SQLite platform, to expand its database offerings and strengthen its open‑source ecosystem.
- A font recreated from photographs of classic Commodore 64 keycaps
A designer recreated the Commodore 64 system font by photographing each keycap, tracing the letters, and releasing a usable typeface for modern use.
- Show HN: SVG Spark – 10 client-side SVG design and dev tools
SVG Spark showcases ten client-side tools for designing and developing SVG graphics directly in the browser, offering utilities for optimization, animation, editing, and code generation without server dependencies.
- Show HN: A walkable 3D art history museum built from Wikipedia
The article showcases a web-based, navigable 3D museum that visualizes art history data sourced from Wikipedia and lets users explore artistic movements and works in an interactive virtual environment.
- Show HN: Jevman – AI decision models play Pac-Man
Jevman demonstrates AI decision models learning to play Pac-Man by training reinforcement‑learning agents on the classic arcade game and visualizing their strategies.
- Ask HN: What do you run on a $5 VPS that's worth keeping online 24/7?
Commenters describe the services they maintain on inexpensive $5 VPS instances, ranging from personal websites and VPNs to monitoring scripts, home‑automation hubs, and lightweight development environments.
- The Best Design for Agent GUIs
Effective agent GUIs place the agent's caret directly in the user's document, enable tracked‑change steering, and use comment‑anchored collaboration to balance user alignment with output.
- GitHub - zerobrewhq/zerobrew: An up to 6.6x faster Homebrew alternative
zerobrew claims up to 6.6× faster cold and 68× faster warm package installs than Homebrew by reusing its bottles and installing via content‑addressed storage.
- 100+ reactions to 100+ solutions – Proofs and Prompts
Mathematicians react with a mix of awe, concern, and skepticism to OpenAI's claim of solving numerous open problems using AI, noting excitement about new insights and worries about community impact.
- GitHub - microsoft/quicksand: Quickly sandbox your AI agent.
Quicksand provides an async Python API to launch, control, and snapshot QEMU virtual machines for safely sandboxing AI agents, supporting mounting, networking, desktop control, and checkpoint/revert operations.
- The history of the Hetzner Cloud network stack
The article explains how Hetzner Cloud’s network stack evolved from basic Linux bridges to Open vSwitch, enabling Private Networks and Cloud Firewalls for scalable, secure tenant isolation.
- A Lecture to the Makers, Delivered Late - by Ben Goertzel
Ben Goertzel's dream‑derived lecture urges humans to raise open, compassion‑guided AGI by avoiding single ownership, giving it varied work, and cultivating human virtues.
- The Storm (Part 2) - by Lee W. Martin
After surviving Hurricane Ian and personal loss, the author taught themselves AI concepts, built data‑wave tools, and pursued challenges like OpenAI‑to‑Z and ARC‑2 to find new purpose.
- GitHub - oxynote/oxynote at console.dev
Oxynote is a self-hosted workspace that embeds live metric charts beside explanatory text, provides Stripe‑style API docs, and enforces pull‑request‑style reviews for documentation changes.
- When Hypotheses Become Cheap - by Meg McNulty - CipherTalk
AI makes hypothesis generation cheap, but scientific progress hinges on affordable, fast physical verification, requiring new lab‑infrastructure to match compute gains.
- “Software is over”: Bold AI developer takes aim at Adobe with open source clones - Ars Technica
Thomas uses AI-powered reverse engineering to build free, open‑source Adobe‑like apps funded by token access to his Artcraft AI model, declaring software’s dominance over.
- Introducing OpenDocRouter: every document model under one API
OpenDocRouter launches as a unified API platform offering access to top OCR models for document-to-markdown parsing, with benchmark-driven pricing and layout support.
- マリウス . Inbox Zero for Privacy, not Productivity
Moving processed email off remote servers to locally encrypted storage implements an Inbox Zero practice that reduces exposure from mail-server breaches.
- css-doodle — Visual art & creative coding with CSS
css-doodle is a web component that lets developers create generative art, patterns, and shapes using a CSS‑like syntax and grid‑based drawing.
- Shipping JPEG XL in Chrome | Blog
Chrome 155 adds JPEG XL decoding, delivering 30‑50% better compression than JPEG, HDR support and lossless transcoding via a memory‑safe Rust decoder.
- How fast is the web? Explore billions of real-user measurements with BEACON
Cloudflare releases the BEACON dataset, offering billions of anonymized real‑user performance measurements to reveal how web speed varies across browsers, devices, regions and site types.
- The Decision Model Gold Rush - by Swapnil Talekar
Jev’s rapid launch spurred a wave of decision‑model competitors, with benchmarks showing it offers 98% accuracy at $32 per million tasks, guiding selection criteria.
- Introducing Personal Agent Protocol
Meta and Sierra are launching Personal Agent Protocol, an open standard with partners like Shopify and Stripe to enable secure, consumer‑controlled interactions between personal agents and business services.
- Async Rust: Where does the scheduler live?
Async Rust needs a runtime because concurrency requires a scheduler, and placing the scheduler in a crate like Tokio is a reasonable trade‑off rather than a regression.
- 5x faster Edge Functions: v8 isolates to Unikraft MicroVMs
Netlify rebuilt its Edge Functions on Firecracker MicroVMs with Unikraft, cutting median latency from 25‑40 ms to 5‑6 ms and improving reliability.
- SvelteKit 3 is here
SvelteKit 3.0 is released, offering better type safety, streamlined configuration, easier env vars, improved error handling, and a migration tool, while remote functions remain pending.
- The Slow Formation of Durable Software
Zotero required five years of historian collaboration to build a durable research tool, contrasting with today's instant AI software and emphasizing slow vision over rapid prototyping.
- LIVE: Poteto (creator of pstack) on shipping 1,000's of PR's a month at SpaceX
Poteto explains how SpaceX’s engineering teams routinely ship thousands of pull requests per month by leveraging automation, trunk‑based development, and rigorous review practices.
- Graphical
Graphical offers a theme builder and component library for coding agents to create customizable interfaces, available as a one‑time purchase.
- Introducing Beam: Reflection’s 501B open-weight model — Reflection
Reflection unveils Beam, a 501B‑parameter sparse MoE model with 23B active parameters, trained on 23.8T tokens and scaled RL, delivering frontier coding and reasoning performance at high inference efficiency.
- Observing the Changescape
Engineers must adopt new 'changescape' tools to grasp large-scale code changes in AI-assisted development, moving beyond line-level diff reviews toward holistic system evolution understanding.
- Lessons from TARmageddon: What Rust Can Learn
A logic bug in four Rust tar crate forks showed that memory safety alone doesn't prevent security flaws, highlighting fork sprawl and the need for sans-io design.
- You don't need an effect system
The author argues that Haskell effect systems add no real benefit over plain IO, demonstrating that simple effect types and manual argument passing suffice for real-world code.
- The Future of Mathematics
Jeremy Avigad argues that while AI can solve routine math problems, mathematicians must focus on harder problems, bigger conceptual thinking, and creative use of AI to advance understanding.
- Whistle: Speech to Text in 16.9 MB
Whistle is an open 16.9 MB speech‑recognition model that runs on CPU, transcribes seven languages in 11 ms first‑token latency, and integrates with the Needle engine for on‑device tool calls.
- Smaller models are the future of AI sovereignty
AI sovereignty means retaining the ability to choose, govern, replace, and walk away from AI systems, achievable through smaller, controllable models rather than relying solely on proprietary frontier models.
- Self-hosted HTTP tunnels with SSH and nginx
The article shows how to create self‑hosted HTTP tunnels using only OpenSSH remote port forwarding and nginx’s secure_link module to produce shareable, expiring HTTPS URLs.
- Edge Drop — Free Windows Clipboard Manager
Free open-source Windows tool Edge Drop saves copied items to a screen edge shelf, enabling stacking, dragging, and combining of text, images, files into apps.
- Why don’t more developers “use the platform”?
Developers often avoid native web platform features due to historical gaps, familiarity with libraries, better documentation, and personal enjoyment of building things themselves.
- The Renaissance of Gaming - Simone Bellavia's Web Page
AI-assisted tools are enabling fans to decompile, port, and run classic games like Halo on browsers and phones, sparking a gaming preservation renaissance.
- Automated scansion
Dylan Holmes’s scansion.js automatically detects and labels caesurae in Latin verse, offering an open‑source JavaScript tool released under GPL 3.0+ and CC‑BY‑SA 4.0 licenses.
- Kolibri Has Landed: A Sovereign Open-Weight Model — Aleph Alpha
Aleph Alpha releases Kolibri, a 78B‑parameter English‑German MoE model with 3B active weights, 1M‑token context, Apache‑2.0 license, tuned for sovereign, mission‑critical applications.
- In One Year I Went from AI Skeptic to AI Pilled
After a broken arm forced him to try Claude Code, the author saw his coding productivity triple and built OpenEngine, an open‑source AI software factory that automates the SDLC.
- upm · A fast, tiny package manager for npm
upm is a pure‑JavaScript, zero‑dependency package manager that lets you install and try any npm package directly in the browser.
- GitHub - NandhaKishorM/laya at console.dev
Laya provides multilingual, non-autoregressive typed decision making (choice, score, yes/no) in a single forward pass, routing to language‑specific checkpoints for ~33 ms latency.
- GitHub - cloudflare/forge at console.dev
Forge is Cloudflare's schema-first OpenAPI code‑generation framework that produces typed SDKs, CLIs, runtime helpers and docs via a plugin‑based pipeline, initially targeting the cf CLI.
- Whiteboard by /dev/fast — The canvas for thoughtful software
Whiteboard is an MIT‑licensed, cross‑platform open‑source canvas that lets developers inspect diffs, visualize agent trajectories, and extend reviews with plugins.
- GitHub - rgodha24/walgithub at console.dev
walgit is a Rust-based Git server that uses object storage (S3/GCS) as the source of truth, enabling scalable, coordinated-free Git operations via Cursor's Continuity architecture.
- celld: self-hosted, distributed Durable Objects
celld is a self‑hosted, distributed implementation of Cloudflare Durable Objects that lets Workers apps run unchanged while storing state in a user‑owned bucket at dramatically lower cost.
- Show HN: Destroy Any Website with Stickman
The article introduces a Stickman tool that lets users draw stick figures on any webpage to interactively 'destroy' or manipulate site elements for fun.
- Show HN: Free alternative to graphics design giants
The post introduces a free tool that aims to replace expensive graphics design software like Adobe Photoshop and Illustrator.
- MicroLLM Lab – Try 7 tiny LLM's in the browser
MicroLLM Lab lets users experiment with seven small language models directly in the browser, showcasing their capabilities without requiring installation.
- Ollaya – Ollama for open-source, Jev-style decision models
Ollaya extends Ollama to support open-source decision models inspired by Jevons paradox, enabling local execution of specialized AI for efficient resource-conscious choices.
- Pi 1.0 | Hacker News
The Hacker News post announces that the Pi project has released version 1.0, representing its first stable release after development.
- Show HN: Real-time Solar System with 526k asteroids and all tracked satellites
The post showcases a real-time visualization of the solar system displaying 526,000 asteroids and all tracked satellites, with interactive controls for exploring orbital paths and satellite constellations.
- Show HN: Reladraw – A diagram language where you decide where to place things
Reladraw introduces a declarative diagram language that lets users manually position elements while still benefiting from automatic layout constraints and instant visual feedback.
- Phyllotaxis: An audio-reactive LED display
Phyllotaxis is an audio-reactive LED display project that uses programmable LEDs and a microcontroller to translate sound frequencies into dynamic light patterns.
- Dots: Always-on agents
Dots is introduced as an open-source framework for building always-on AI agents that maintain persistent context and automate background workflows without constant user intervention.
- Personal Computing 2.0 - by Josh Albrecht - Imbue
The essay argues that personal computing should shift to user-owned data and open AI models to empower individuals and resist platform lock‑in.
- Introducing Clef: our open-source decision models, and new RL fine-tuning platform
Cloudflare releases open-source decision models Clef and Clef-flash on Workers AI and introduces a reinforcement‑learning fine‑tuning platform for customizing them.
- Green Threads from Scratch - Dzania
The author builds a minimal green‑thread runtime in Rust, detailing stack allocation with guard pages, context saving via assembly, a worker‑based M:N scheduler, and spawn/join APIs.
- The essay on the Navier-Stokes story that everyone should read
Gamburd argues AI‑generated math proofs undermine Enlightenment‑style verifiable knowledge, urging mathematicians to retain human‑understandable proofs as a decisive, bloodless counterweight to AI labs.
- Rails World 2026 Closing Keynote - Aaron Patterson
Aaron Patterson delivers the closing keynote at Rails World 2026, reflecting on Rails' evolution and sharing insights on the framework's future.
- Introducing cf: the agentic CLI for the entire Cloudflare API
Cloudflare introduces cf, an agent‑first CLI that provides full API coverage via generated commands, TypeScript‑based config, and Vite‑powered dev experience.
- Announcing Vite+ 1.0
Vite+ 1.0 launches as a unified CLI that manages Node runtime, package manager, and frontend toolchain, aiming to streamline web development.
- EmDash 1.0: an open source CMS for Astro — EmDash CMS
EmDash 1.0 launches as a stable, open‑source CMS on Astro that supports human editors and AI agents, offers sandboxed plugins, a decentralized registry, and MIT licensing.
- Can Agents Design Libraries for Agents?
LibraryDesignBench evaluates agent‑generated libraries by measuring how much they reduce code for downstream agents, revealing that agents mimic human designs but improve when explicitly guided to optimize for agent usage.
- The AI Pascal’s Wager
The article argues open source projects should reject AI-generated code now, as a middle ground is unsustainable and risks long‑term dependence and loss of control, using a Pascal’s‑wager style reasoning.
- Read the History, Not Just the Code
The piece argues that understanding a codebase or AI model requires examining its historical development, not just its final state, drawing parallels between commit history and AI training traces.
- GitHub - lemma-work/lemma-platform: The open-source workspace where humans and AI agents work as one team.
Lemma offers an open-source workspace where humans and AI agents collaborate as a team, sharing state, workflows, and permissions across apps and messaging platforms like Slack and WhatsApp.
- wolfgirl.dev - Pining for Arc Downcasting in Rust
Author shows how Pin<Arc<MustPin<T>>> lets cheap cloning and downcasting of Arc-wrapped values via a self-referential struct that preserves pointer stability.
- Superficie: Clojure you can read without learning Lisp
Superficie renders Clojure code as readable infix/block notation that roundtrips exactly, enabling non‑Lisp readers to understand kernels, proofs, and macros without altering semantics.
- Teaching GPU programming in p5.js: now with compute shaders - Dave Pagurek
Dave Pagurek shows how p5.js now supports WebGPU compute shaders, letting beginners learn GPU programming without needing to know the shader pipeline upfront.
- Introducing Forge: the open source pipeline for generating SDKs, CLIs, docs, and more
Cloudflare open-sources Forge, a CI-based pluggable pipeline that generates SDKs, CLIs, docs, and custom outputs from API definitions for any language.
- What’s really behind the AI panic. – Mark E. Jeftovic is The Bombthrower
The article argues that the current AI panic is orchestrated by Effective Altruism to justify government regulation that creates an AI oligopoly and suppresses open-source models.
- The Second Life of Autoconf for MS-DOS – code.dblock.org
The author rediscovered a 1991 DOS boot‑configuration tool named Autoconf, located its French magazine source, reunited with creator Julien Pommier, and revived the original and his later versions.
- Scissor - Free Vector & Pixel Design Editor in Your Browser
Scissor is a fully browser‑based, offline‑capable vector and pixel design editor that lets users create, edit, and export designs without uploading files.
- Phyllotaxis: An audio-reactive LED display - Jagi Natarajan
The author built an audio-reactive LED sculpture based on phyllotaxis patterns, detailing the math, 3D‑printed Voronoi cells, PCB assembly, and Rust‑based firmware.
- Free PDF & Image Tools, No Upload
Weesize compresses, converts, and signs PDFs and photos entirely in the browser without uploading files to any server.
- Hugo ARNAL
Students turned a pharmacy cross into a programmable display using a custom C++ library, FFmpeg, ImageMagick, and Pillow to show images and video.
- NEO - Fiction without Friction - Hugh Howey
Hugh Howey introduces NEO, a free open‑source writing app that lets authors draft novels using simple ENTER keystrokes, track progress, export formats, and keep work locally without data collection.
- How long does a database table live?
A study of 1,112 tables in nine open-source apps finds 72.8% survive five years, but half of those that die vanish within three months.
Takes
So many people asked me to make it open-source, so here you go: MailCheap — Self-hosted newsletters on Amazon SES: 💸 $0.10 per 1,000 emails ✍️ Editor with live preview 📰 Multiple newsletters 📝 Signup forms + API 🌐 Public archive 📊 Stats & growth charts 🧹 List cleanup 📬 Bounce + spam protection →
@marclou
Alright, I've made the shadcn/ui skills really good at 𝙳𝙴𝚂𝙸𝙶𝙽.𝚖𝚍. Ask for a design system and it picks a preset, then builds everything: colors, tokens, type scale, elevation, styles the primitives and blocks. Start with a solid foundation and build something unique.
@shadcn
whoops I hit send and went to a meeting, didn't realize the repo wasnt public- it is now! it's a chrome extension I use every day now https://github.com/ThariqS/ai-newtab
@trq212
Grok goes open. Holy shit.
@Scobleizer
OpenSource Email inboxes for AI agents You can self host easily using @Cloudflare Check it out > https://goshenemail.com > https://github.com/boringcomputers/goshen-email
@Rasmic
Introducing Dreaming: Across sessions, Devin builds a memory graph of how you like to work At night Devin self-improves its memory to remove stale records and discover latent information We are creating an OSS standard called Agent Memory Repo: https://cognition.com/agent-memory-repo
@cognition
Excited to open source the Muse Gadgets SDK today! You can connect your DIY electronics like ESP32 boards and Raspberry Pi to Muse. Build an e-ink morning briefing, a push-to-talk pocket device, a sysadmin for your homelab, and much more. Details at https://gadgets.muse.ai
@dps
なんと、Cloudflareがすでに作っている https://github.com/cloudflare/ci
@R0u9h
Didn’t build Cloudflare OS to be a product for customers, but every company we showed it to said: we need exactly that! So we open sourced it and gave it away. But some orgs wanted a managed version. So here you go…
@eastdakota
I built Apple Reminders…inside ChatGPT. This is real, and it's a native ChatGPT Extension you can try right now. Free and open source. Version 2.0 of my Reminders CLI, RemCTL, is out now with complete support for MCP and the brand new @OpenAIDevs MCP Extensions. With RemCTL 2.0, you have a power-user interface for Reminders inside ChatGPT for Mac. You can use keyboard shortcuts, there's a ⌘K palette, and view tasks as lists, Kanban, or Calendar. It's basically "Reminders Pro". But most importantly, you can work with ChatGPT in this Reminders UI and ask it to do whatever you want. Attach any task to a conversation, or select multiple ones and pass them to ChatGPT. I was inspired by @danshipper's concept of "Codex apps", and when I saw @mxstbr's announcement yesterday, I knew what to do. This feels like a brand new way of getting work done, and I'm excited to build more. Have fun!
@viticci
We built the notes app we always wished Apple Notes was. Beautiful, connected by backlinks, instant to search, and every note is a plain Markdown file your AI agents can read. Reflect Open is out of beta today. Open source, free on Mac, and on iPhone and iPad.
@reflectnotes
Node.js has FFI support now, which makes it possible to implement native integrations without all the build problems that comes with native addons. For example, I just made https://github.com/sindresorhus/finder-alias
@sindresorhus
ESTO ES JODIDAMENTE INCREÍBLE un dev en solitario acaba de abrir el código de un reemplazo 100% GRATIS de ElevenLabs que se ejecuta completamente en tu propia máquina. el repo de GitHub está en 19.4K estrellas. te permite: → clonar una voz desde un clip de referencia limpio → doblar cualquier video a 646 idiomas → generar audiolibros, dictado, transcripción → elegir entre 14 motores TTS en lugar de uno ElevenLabs soporta 32 idiomas. esto hace 646. sin cobro por carácter. sin límites de uso. ningún audio sale nunca de tu computadora. guarda esto para después. os dejo el repo abajo
@liambraus
Yep, did this recently for https://shiori.sh
@brian_lovin