📦 lajosdeme/mole

Mole

Mole is the deep research assistant that keeps your artificial intelligence queries honest and strictly on budget. Unlike standard chat search engines that can burn through cash on runaway queries, this clever tool lets you set a hard financial ceiling so your runs stop the moment they hit your limit. It decomposes your questions, reads sources, and verifies every single claim against a verbatim quote before writing an answer. If you want to analyze sensitive local data, it keeps your spreadsheets private on your machine, rendering local database queries so only aggregate statistics ever leave your system.

📰 https://news.ycombinator.com/item?id=49303046

🆔 @hackernewsgithubprojects

📦 wmww/tawc

tawc

Run full desktop Linux apps right on your Android device without rooting your phone. This is possible thanks to a project called tawc, which stands for Tess's Android Wayland Compositor. It sets up a complete environment directly on your mobile device, letting you install desktop distributions like Arch or Debian. Instead of sluggish emulation, it bridges your hardware directly to the software, meaning your apps run with smooth, accelerated graphics. You can launch full terminal tools or graphical desktop programs right from your app drawer. It is a brilliant way to turn your everyday phone into a portable workstation.

📰 https://news.ycombinator.com/item?id=49313068

🆔 @hackernewsgithubprojects

📦 thelastoutpostworkshop/espconnect

espconnect

espconnect is the browser-based control center that finally lets you manage your microcontrollers without installing any bloated desktop software. Just plug in your chip, open your browser, and instantly access a complete suite of diagnostic and management tools. You can inspect hardware details, manage files directly on your device, and even flash new firmware right from a tab. It supports common filesystems like LittleFS and features an experimental inspector to let you peek inside non-volatile storage. It is the ultimate plug-and-play web workspace for makers who want to skip the setup and get straight to building.

📰 https://news.ycombinator.com/item?id=49313675

🆔 @hackernewsgithubprojects

📦 harshalgajjar/reminal

Control Any Computer From Your Browser with Reminal

Control any of your computers directly from a web browser without dealing with SSH keys, firewalls, or open ports. Running a single command on your machine generates a secure session link and a QR code. When you scan it with your phone, you instantly get a live, responsive terminal right in your browser. Beyond the command line, you can stream and fully control individual application windows or mirror your entire desktop. It even lets you share local development ports, transfer files between machines using simple one-time codes, and control your Mac with the lid closed by automatically spinning up a virtual display.

📰 https://news.ycombinator.com/item?id=48747739

🆔 @hackernewsgithubprojects

📦 mhenry3164/twenty-crm-mcp-server

twenty-crm-mcp-server

Manage your customer relationships directly through your favorite artificial intelligence assistant by hooking up your customer relationship manager. By running this clever integration, you can perform full create, read, update, and delete actions on your people, companies, opportunities, and tasks using plain natural language with Claude. It completely solves the headache of navigating complex customer databases by mapping flat parameters to composite database fields automatically and handling actual cursor-based pagination behind the scenes. It is perfect for developers who want to quickly query, search, and update their customer pipelines without leaving their chat terminal. Give it a spin to see how easily your workspace data comes alive.

🆔 @hackernewsgithubprojects

📦 studio-dots-ai/dots3-note-prev

dots3-note-prev

Process text, images, videos, and full audio files all at once using a highly efficient open-weight artificial intelligence model. The dots3-note-prev repository introduces a massive mixture of experts model that boasts two hundred and eighty billion total parameters while keeping only sixteen billion active at any given time to save computing power. The most exciting part is its massive support for up to five hundred and twelve thousand tokens of context, allowing you to feed in entire documents or lengthy media clips without losing the thread. It is a highly capable, lightweight visual and audio intelligence tool ready to run on standard pipelines.

🆔 @hackernewsgithubprojects

📦 wheelry/deep-skill-finder

Deep Skill Finder

Most AI agents pick tools based on the creators' hyped-up descriptions, which means they easily break on real-world tasks. That is why deep-skill-finder is such a game-changer for autonomous workflows. Instead of relying on raw keyword matching or meaningless download counts, this clever tool crawls the web and matches the perfect skill for your agent based on actual execution history. It queries community run records and benchmark reports to see what has actually run successfully before recommending or installing anything. It essentially gives your agent a built-in detector for broken dependencies, so it can solve your task without falling into dependency hell.

🆔 @hackernewsgithubprojects

📦 pengchujin/tower

Tower

The open source iOS app Tower lets you safely manage all your internet proxy subscriptions and custom nodes entirely on your device without ever uploading your private credentials to a third-party server. Built natively with Swift UI, it parses your custom node links and automatically identifies their geographic locations using an offline, built-in IP database rather than calling external services. It then lets you bundle and convert these nodes into precise local configurations for seven major clients like Surge, Clash, and Shadowrocket. This keeps your credentials secure while giving you custom, fast routing. Grab this native tool on GitHub to reclaim your network privacy.

🆔 @hackernewsgithubprojects

📦 5p00kyy/club-5060ti

Club 5060ti

Nvidia Blackwell RTX 5060 Ti graphics cards can actually run massive thirty-five billion parameter local language models at a stunning one hundred and thirty-one thousand context window. A clever open-source repository called club-5060ti provides exact, copyable system presets and verified benchmarks to maximize local artificial intelligence on these specific sixteen-gigabyte cards. Instead of guessing your settings, it gives you highly optimized, real-world configurations for single, dual, and multi-gpu configurations, proving exactly what works without the usual guesswork. It is the ultimate playbook to get every ounce of local intelligence out of your hardware.

🆔 @hackernewsgithubprojects

📦 holtwick/bx-mac

bx-mac

Your AI coding assistants have broad system access, meaning a single hallucinated command or misguided tool call could easily read your private photos, tax documents, or SSH keys. The bx-mac tool instantly wraps any application in the macOS sandbox, blocking access to everything on your machine except the exact project directories you specify. It works dynamically without containers or slow virtual machines, so tools like VS Code, Claude Code, and your terminal run at full speed while keeping personal folders entirely off-limits. Just run bx with your folder path, and you get an instant safety net that protects your private data while keeping your development workflow completely functional.

📰 https://news.ycombinator.com/item?id=47674056

🆔 @hackernewsgithubprojects

📦 hughhowey/neo

Neo

Write your next book without any distractions using a minimal word processor designed specifically for novelists. Instead of treating your manuscript like a generic document, this tool formats your workspace to look and feel like a real book, complete with automatic chapter numbering and drop caps. It organizes your projects on a visual bookshelf and even lets you tuck away cut passages into a virtual graveyard to restore later. When you need to force yourself to write, you can enter a locked fullscreen focus mode that literally won't let you leave until you type your way out. It is a wonderfully simple, free way to focus entirely on your storytelling.

📰 https://news.ycombinator.com/item?id=49314647

🆔 @hackernewsgithubprojects

📦 aaronbeckmann/esp-dashboardplus

ESP-DashboardPlus

ESP-DashboardPlus is the on-device library that finally makes building sleek web dashboards for your microcontrollers incredibly simple. Instead of writing messy HTML or struggling with slow interfaces, you can drop this C plus plus library onto your ESP32 to instantly serve a gorgeous, responsive interface directly from the chip. It packs sixteen different interactive cards, from live charts and gauges to color pickers and sliders, all updating instantly over fast web sockets. You even get a built-in terminal console and drag-and-drop wireless firmware updates right out of the box. It is the ultimate way to control your hardware projects in style.

🆔 @hackernewsgithubprojects

📦 playform/aphrodite

Aphrodite LLM Context Compressor

Massive terminal outputs, raw code dumps, and bloated build logs are constantly eating your AI agent's budget and ruining its focus. Aphrodite fixes this by compressing massive developer outputs down to a microscopic scale. Instead of sending thousands of raw tokens to your large language model, it instantly intercepts the noise, runs a lightning-fast analysis, and replaces it with a tiny fifteen token preview. The model reads the structured preview instantly and only fetches the full text if it actually needs it. It saves you tons of tokens while keeping your AI incredibly fast, smart, and context-aware.

🆔 @hackernewsgithubprojects

📦 wheelry/meyo-skill-finder

meyo-skill-finder

meyo-skill-finder is the smart search utility that helps your AI agent find and install the absolute best tool for any specific task. Instead of relying on vague marketplace descriptions or star ratings, this tool indexes AI skills from across the web and ranks them using actual community performance data and execution records. When your agent faces a tricky task, it translates your natural language request into a semantic search, scans real-world test logs, warns you of potential bugs before installation, and automatically sets up the verified winner. It is a fantastic way to ensure your agent always runs on proven, working tools.

🆔 @hackernewsgithubprojects

📦 klipitkas/tunnl.gg

Tunnl.gg

Expose your local development servers to the public internet instantly using a single built-in terminal command. By leveraging the secure shell protocol already installed on almost every computer, this lightweight tool lets you share your work without installing any custom client software, registering for accounts, or configuring complex settings. It automatically provisions a memorable, randomly generated subdomain with full Let's Encrypt encryption, handles modern websocket connections, and includes smart safeguards like automatic rate limiting and browser phishing warnings to prevent abuse. It is the ultimate self-hosted alternative for sharing local progress with clients and testing webhooks on the fly.

🆔 @hackernewsgithubprojects

📦 tempestai-dev/tempest

Tempest: Run 5 AI Agents Without Burning Cash

Five AI coding agents can work on your project at once without stepping on each other's toes. That is the core idea behind Tempest. Instead of every agent spending precious tokens reading your files from scratch, Tempest builds a shared map of your codebase. This allows the agents to pull information from a shared graph rather than scanning your files repeatedly. The result is massive savings, cutting token usage by up to eighty-six percent in benchmarks. You can run multiple agents in parallel, each in its own safe space, and review their work easily. It is a smart way to get more done with less cost.

🆔 @hackernewsgithubprojects

📦 hiroppy/tmux-agent-sidebar

tmux-agent-sidebar

Running multiple AI coding agents at once usually turns your terminal into a guessing game, but tmux-agent-sidebar fixes that by giving you a single dashboard for all of them. It sits as a simple sidebar inside Tmux, watching every Claude Code, Codex, and OpenCode pane across your entire setup in real time. Instead of constantly switching windows to check if a job finished or got stuck, this tool reads the live activity from each agent and displays their current status, recent prompts, and Git changes in one clear view.

🆔 @hackernewsgithubprojects

📦 kavinsood/yaos

YAOS: Real-Time Obsidian Sync Without the Terminal

YAOS makes Obsidian notes appear on every device the instant you type them, without needing a terminal, a database, or a paid subscription. It runs on a free Cloudflare Worker that you deploy with a single click, turning your personal infrastructure into a live sync engine. The standout feature is its use of CRDTs to merge edits in real time, so two people typing in the same note never get stuck with conflicting copies. It works entirely locally, keeping your vault as plain Markdown files while syncing changes instantly across desktop and mobile, even offline, and integrates cleanly with scripts or AI agents that edit files directly on disk.

🆔 @hackernewsgithubprojects

📦 docwriter-org/plain-writing-skill

Plain Writing Skill: Fixing AI's Wordy Output

Plain Writing Skill is the instruction set that forces AI to cut the corporate fluff and speak like a human. It acts as a strict editor for your chatbot, stripping away jargon, empty buzzwords, and confusing sentence structures from its responses. Instead of letting an AI bury the good news in a paragraph of marketing-speak, this tool ensures the main point is delivered first and clearly. It is essentially a quality control gate that makes sure your code assistant actually explains its work in simple, everyday language. It is a clever way to keep your AI honest and readable.

🆔 @hackernewsgithubprojects

📦 rhoban/microban

Microban: The Robot You Can Print at Home

Microban is a thirty centimeter tall humanoid robot that you can print entirely on your desktop 3D printer. For around six hundred dollars, this open source creation gives you a fully functional body with nineteen servo motors and a tiny computer brain. It is built specifically for students and hobbyists who want to understand how bipedal walking works without buying expensive commercial hardware. Because the design is free to modify, you can experiment with movement algorithms, mechanical tweaks, and software upgrades directly on the physical unit. The goal is to make advanced robotics research accessible to anyone with a standard printer and basic tools.

🆔 @hackernewsgithubprojects