📦 getopenpost/openpost

One draft, every platform: OpenPost

Turn one idea into ready-to-post content for every social network you manage. OpenPost is a self-hosted workspace where you write a single draft, then tweak the text and timing for each channel before publishing. It solves the tedious task of rewriting the same update for LinkedIn, X, and Mastodon, ensuring your voice stays consistent everywhere. The standout feature is its shared draft system, which lets you adjust media and scheduling for each destination while keeping a master version for your records. You can even run the whole stack on your own server with Docker. If you want to stop copying and pasting between apps, this is your new home base.

📰 https://news.ycombinator.com/item?id=49626154

🆔 @hackernewsgithubprojects

📦 kamaienterprises/aec-geometric-bench

How Good Are AI Models At Reading Blueprints?

AEC Geometric Bench lets you see exactly how well artificial intelligence actually reads construction blueprints. Most models can spot a room's general shape, but they fail to identify specific doors, windows, or fixtures needed to price a build. This project tests six systems on real architectural sheets using the same strict rules. The results are surprising: the specialized tool reads the drawings accurately, while the biggest general models miss most of the critical details. If you are building tools for construction, this open dataset gives you the real numbers to back up your claims, proving that general intelligence alone is not enough to handle precise technical documents.

🆔 @hackernewsgithubprojects

📦 tigerless-labs/agent-memory

Agent-Memory: AI That Actually Remembers

AI coding tools usually forget everything the moment you close the session, but agent-memory fixes that by keeping your data as simple text files instead of hidden databases. This means you can always open them and read exactly what your AI learned, without needing any special software or online accounts. It works locally on your computer, so different tools can share the same memory instantly. Because the data is just plain words, you can delete the search index at any time without losing your notes, and you can even edit the memories yourself with a standard text editor.

🆔 @hackernewsgithubprojects

📦 innate-labs/noobi.ai

Noobi.ai: The Desktop Agent That Codes Your Games

Noobi.ai is the local-first desktop agent that actually turns a simple text prompt into a fully playable browser game. It’s not just a code generator; it’s a complete development loop running right on your machine. You type what you want, and the agent builds, reviews, and compiles the game locally using a specialized app server. The best part? It’s private. Your prompts and code never leave your computer, which is a massive win for anyone worried about data privacy. It takes the friction out of coding by handling the heavy lifting of syntax and logic so you can just play.

🆔 @hackernewsgithubprojects

📦 primefoldtools/andon

Andon: Stop AI Mistakes From Repeating

Andon is the quality control system that actually stops your AI from repeating the same errors. Most tools just try to make the output sound better, but Andon borrows a hard lesson from factory floors to make sure a mistake never happens twice. If your agent claims a task is finished without actually testing it, or forgets what you agreed on last week, this is the fix. It uses a simple hook to catch those false claims and logs the defect so the system learns from it. It is not just a memory file; it is a discipline that turns every mistake into a permanent guard.

🆔 @hackernewsgithubprojects

📦 docker/sbx-kits-contrib

sbx-kits-contrib

Docker Sandboxes lets you lock down AI agents in isolated environments, but community kits let you pre-load specific tools, network rules, and even entire custom operating systems before the agent starts. The sbx-kits-contrib repository is a growing collection of these ready-made packages, each defined by a single declarative file that tells the sandbox exactly what to install and which network connections are allowed. This solves the problem of manually configuring complex, secure environments for every new project. The most surprising part is how granular the control is, allowing you to specify exactly which websites an agent can reach and what files appear in its home directory.

🆔 @hackernewsgithubprojects

📦 iamwavecut/mlx-dlss

MLX DLSS on Apple Silicon

NVIDIA's proprietary neural rendering engine is now running natively on Apple Silicon, bypassing the need for any Windows driver. This project, called MLX DLSS, extracts the mathematical logic directly from standard game files to recreate that specific visual quality on Mac hardware. It handles both super-resolution and frame generation, allowing users to process video and images with the same fidelity found in high-end PC gaming. The entire pipeline runs locally using the MLX framework, meaning no external servers or proprietary binaries are required during playback. For developers, this offers a fascinating look at how deep learning models can be reverse engineered and ported across completely different hardware architectures.

🆔 @hackernewsgithubprojects

📦 giuliastro/harness-remote

Harness Remote: Switch AI Coding Agents Without Losing Your Work

Harness Remote lets you start a coding task on one machine and pick it up on another, switching between different A I agents along the way. Most tools force you to copy and paste your work into a new chat when you change tools, which is messy and error-prone. This project solves that by acting as a control layer for your existing agents, like Claude Code or OpenCode, so you can hand off a task while keeping the original context intact. Your code and credentials stay on your own computers, not in some cloud.

🆔 @hackernewsgithubprojects

📦 feizhululu/real-api-pricing

Real API Pricing: The Truth About Your AI Subscription

Real API Pricing calculates the true cost of every AI token by dividing your monthly subscription fee by the actual amount of usable tokens you can burn, solving the mystery of why some expensive plans feel stingy while others feel infinite. It takes the vague marketing numbers from major AI providers and turns them into a single, comparable dollar-per-million-token metric using a standardized workload that accounts for how modern coding agents actually use memory and context.

🆔 @hackernewsgithubprojects

📦 nex-agi/nex-n2.5

Nex-N2.5: The AI That Sees

Nex-N2.5 is a family of AI models that actually look at your screen to work, not just read text. The biggest surprise? Vision is now its main way of checking its own work. It acts like a digital worker that clicks buttons, browses the web, and even runs code, constantly glancing at the results to fix mistakes on the fly. This is huge because it means the AI can handle long, messy real-world tasks where things go wrong and need correcting. It is less like a chatbot and more like a tireless junior employee who actually watches what it is doing.

🆔 @hackernewsgithubprojects

📦 karnyadavdev/fontwizard

Font Wizard: Change Your Windows System Font

Make your Windows look exactly how you want it to with Font Wizard. This tool solves a frustrating problem: most customization tools only change the font for older programs, leaving modern apps and the taskbar stuck with the default look. Font Wizard fixes that by applying your chosen typeface to every single part of the system, including modern Windows apps, browser engines, and system components, so your desktop has a truly unified aesthetic. You just pick a file and apply it, and it handles the technical complexity in the background, even ensuring your icons stay readable.

📰 https://news.ycombinator.com/item?id=49537066

🆔 @hackernewsgithubprojects

📦 bradegithub/dsh-plugins-marketplace

The One-Click DSH Plugin Marketplace

The DSH Plugin Marketplace is the web interface that finally turns thousands of scattered GitHub repositories into a single, manageable catalog for your DeepSeek Harness setup. Instead of hunting for add-ons, you open the settings page to browse a curated list of every compatible plugin automatically. With one click, you can install, update, or remove these tools without ever touching a command line. It even pauses if a plugin needs an API key, so you never have to guess. This makes it incredibly easy to discover and manage the best extensions for your AI workflow.

🆔 @hackernewsgithubprojects

📦 waynchi/gamedevbench

GameDevBench: Can AI Code Games?

GameDevBench puts coding agents to the ultimate test by having them build real games inside the Godot engine. Instead of simple text puzzles, this benchmark throws thirty-three three complex challenges at A.I. models, forcing them to handle sprites, shaders, and visual logic simultaneously. It turns out that even the smartest agents struggle, solving only about sixty-four percent of these tasks. This project fills a huge gap in testing how well A.I. understands visual and interactive software. By measuring success against real game development standards, GameDevBench reveals just how far coding agents still have to go. It is a fascinating look at the true limits of artificial intelligence in creative coding.

🆔 @hackernewsgithubprojects

📦 heroesjouney/aimoviestudiov2

AI Movie Studio 2

AI Movie Studio 2 is the browser-based filmmaking workstation that lets you direct entire scenes in a virtual 3D space. Built by a veteran camera department lead, it solves a critical problem for independent creators: expensive production costs. You don’t just generate images; you place cameras, frame shots with cinematic presets, and use AI to maintain character consistency across every frame. The most surprising feature is the continuity system, which uses reference images to ensure your actors and locations look identical in every shot, preventing the uncanny valley effect that plagues standard AI video.

🆔 @hackernewsgithubprojects

📦 openai/navierstokesandeuler

Navier Stokes and Euler: Proving Fluids Can Break

Navier Stokes and Euler is the proof repository that proves fluids can actually break apart. Instead of just showing these results on a whiteboard, the team used a computer math system to check every single logical step to make sure the answer is perfect. It tackles a huge mystery in science where we prove that under certain conditions, water or air flows can rip apart in finite time. This isn't just a simulation; it is a verified fact that holds up to the toughest scrutiny. It shows us that the equations describing how fluids move can fail in very specific ways.

🆔 @hackernewsgithubprojects

📦 danilaa1/drawably

Drawably: UI that draws itself

Make your app feel like a hand-drawn sketch. Drawably wraps standard buttons, checkboxes, and text boxes in a wobbly, animated pen stroke that looks like it is actively being doodled by a human. The coolest part is that every single time a control appears on the screen, the library generates a brand new, unique sketch using a bit of controlled randomness. It is incredibly lightweight, with zero dependencies and a tiny footprint, yet it feels alive. The real inputs stay in the background for accessibility, so screen readers and keyboards work perfectly while the visual layer just boils gently in the foreground.

📰 https://news.ycombinator.com/item?id=49630353

🆔 @hackernewsgithubprojects

📦 amirmushichge/video-states-website

Video States Website: The Click-Driven Video Site

Turn a static web page into a cinematic experience by letting visitors click through pre-recorded video transitions instead of scrolling through text. This project provides a complete asset pack and build guide for creating an interactive site where four simple controls—scene, lighting, clothing, and cast—play short, seamless video clips that morph the entire viewport. The most surprising part is that it uses paired forward and reverse footage, meaning when you click a button, the scene changes with a fluid motion, and clicking reset plays that exact same motion backwards to return you to the start.

🆔 @hackernewsgithubprojects

📦 williamliu-1997/gaussian-splat-lite

Render Massive 3D Scans in Your Browser

Gaussian-Splat-Lite drops photorealistic 3D scans straight into your browser without a single server round trip. It solves the classic problem of giant point clouds freezing your screen by letting the graphics card sort and stream millions of data points on the fly. The coolest part is how it handles huge real-world maps. It uses special math to keep coordinates accurate even when the scene is millions of kilometers wide, so buildings and landscapes stay sharp. It also blends depth information perfectly, meaning objects hide behind each other instead of looking like a messy cloud of floating dots.

🆔 @hackernewsgithubprojects

📦 0x65dgerunner/react-grid-loader

React Grid Loader: 130+ Animated Loaders

Turn your boring loading screens into custom pixel art by mixing and matching over a hundred different animation patterns. This open-source tool for React lets you build spinners from a grid of tiny dots that you can reshape, recolor, and animate in ways that are surprisingly creative. Instead of just picking a generic spinner from a library, you can design your own unique loading experience right in your browser and export the code instantly. It is perfect for developers who want their apps to feel polished and personal without spending hours writing complex animation code from scratch.

🆔 @hackernewsgithubprojects

📦 andymatuschak/interactive-transcripts

Interactive Transcripts: Edit Audio Like Text

Your audio files just became editable text inside Obsidian. The plugin called Interactive Transcripts turns raw audio into a living transcript where you can literally delete a sentence and the audio skips that part, just like cutting a video. It runs everything locally on your Mac, so your private meetings stay private while you fix typos or trim rambles. The coolest part is that it keeps the timing in sync, so clicking any word jumps the audio to exactly that moment. This is a massive win for podcasters and researchers who hate switching between text editors and video software.

🆔 @hackernewsgithubprojects

📦 sdli1995/dlssg_for_sm86

DLSSG SM86 for RTX 30 Series

Give your RTX 30 series GPU the same AI frame generation tech that newer cards enjoy. This project, called DLSSG SM86, injects a special file into your games to enable DLSS Frame Generation on hardware that officially lacks support. It works by swapping in a compatible version of the technology so your graphics card can create extra frames between real ones, effectively doubling or even tripling your frame rate. In tests, this turned a sluggish fifty frames per second in Black Myth Wukong into a smooth one hundred and fifty.

🆔 @hackernewsgithubprojects