Watch a Flock of Birds Fly in Your Terminal.This terminal flock animation tool is written in C and simulates flocking with the Boids algorithm. It draws birds in the terminal using Braille dots and supports 3D mode, eagle hunting, birds spelling out text to show a clock, and a screensaver mode.

Run 100B+ Parameter Models Without a GPU.This is a mixture-of-experts inference engine written in pure C, with zero dependencies and no GPU required. It keeps only the parts the model needs on every run in memory, and reads the remaining expert weights from disk on demand. With as little as 8GB of RAM, it can run open-source large models such as GLM, Kimi, DeepSeek, and Qwen, and provides OpenAI- and Anthropic-compatible APIs. It works on Windows, Linux, and macOS.

Turning the Linux Desktop into a Desk.This project is a Wayland-based Linux desktop environment, with the entire room rendered using ray tracing. The desktop is a wooden desk, windows are sheets of paper; when minimized, they fold up like letters, and when closed, they are crumpled into paper balls and tossed away.

A Drag-and-Drop File Format Converter.This is an offline file format converter built in C#. It supports over 200 formats, including images, PDF, Office documents, ebooks, audio/video, archives, and CAD. To use it, just hold Ctrl and drag a file—a format ring appears next to the cursor; drop the file on the target format to convert. Files are never uploaded, and it works on Windows, Linux, and macOS.

Use Screenshots to Tell AI How to Modify the UI.This screenshot annotation tool for AI-assisted coding turns specific UI changes into JSON that AI can read directly. It supports hover-based element detection, point and box annotations, dragging components to adjust layouts, pausing videos for frame-by-frame annotation, plus OCR and long screenshots—all processed locally.

An Encrypted File System Written in Go.This project is a file encryption tool built on Go and FUSE. It mounts an encrypted directory as a readable and writable plaintext directory, encrypting file contents and file names separately. The resulting encrypted directory can be stored in cloud drives like Dropbox or incrementally synced via rsync.

One-Click Model Switching for AI Coding Assistants.This is a Go-based model management tool for AI coding assistants that lets Claude Code use Kimi and Codex use DeepSeek. It uses a local gateway to convert between OpenAI, Anthropic, and Gemini API formats, supports multiple coding assistants and model providers, and can share subscriptions with other assistants, automatically switching accounts when quotas run out.

A Terminal Email Client for Sending and Receiving Emails.A terminal email client built with Go and Bubble Tea that lets you send and receive emails via commands, with support for multiple accounts, Markdown email composition, and a plugin marketplace.

Self-Hosted Passwordless Identity Authentication Service.This project is an OIDC and OAuth 2.0 identity authentication service built in Go, with support for Passkeys instead of passwords. Users do not need to remember passwords—they can simply plug in a hardware key such as a YubiKey to log in.

An Open-Source GIS Platform That Runs in the Browser.This project is a geographic information system built with TypeScript for visualizing, browsing, and analyzing geospatial data while keeping data local. It includes over 1,000 analysis tools that run in the browser via WebAssembly, covering terrain, hydrology, LiDAR, and remote sensing analysis. It can run in the browser, on desktop, mobile, and in Jupyter Notebook, and also provides planetary basemaps such as the Moon and Mars.

Self-Hosted All-in-One Online Toolbox.A React-based collection of online tools for everyday tasks such as images, audio/video, PDF, text, date/time, math, and data processing, with all files processed locally in the browser. It offers image resizing and format conversion, video trimming, PDF splitting and merging, time zone conversion, JSON manipulation, and more.

Fitness Tracking App Where You Own Your Data.A self-hostable fitness and weight tracking app with no subscriptions, ads, or telemetry, deployable with a single Docker command. It includes over 5,000 exercises with animated demonstrations, supports weekly workout scheduling, guided workouts, rest timers between sets, automatic personal best detection, and muscle heatmaps to see training volume by body part. You can also import data from apps like Strong and Hevy, and it offers iPhone, Apple Watch, and Android versions.

A Cloud Computer That Runs in the Browser.This is a desktop operating system that runs in the browser, bringing files, apps, and games together in one place for access anytime, anywhere. It includes built-in apps such as Notepad, Voice Recorder, Spreadsheet, and Camera, and offers developers capabilities including AI, cloud storage, a key-value database, and Serverless Workers. Web apps and games you build can be published to its app store.

One Word, One Picture: An IELTS Vocabulary Learning Tool.This is a free, open-source web app for IELTS vocabulary learning. It runs right in your browser and requires no sign-up. It includes over 7,000 core IELTS words, each with an original illustration, two original example sentences, and both British and American pronunciations. Reviews are scheduled using the FSRS algorithm, and every word you master lights up a star on a 3D vocabulary star map.

Give Your TV an Apple TV-Style Home Screen.This open-source Android TV launcher is inspired by tvOS. It has no ads or recommendation feeds and launches automatically on startup. It works on Chinese TV brands such as Xiaomi, TCL, and Skyworth, as well as TV boxes and projectors. It supports organizing apps, creating folders, hiding system apps, and sorting apps via your phone by scanning a QR code. No root required, and you can restore your original launcher anytime.

Using CarPlay on Android Devices.This is a CarPlay app that runs on Android head units. Once installed on the head unit, it enables wired or wireless CarPlay. It supports Wi-Fi Direct and in-car hotspots, can project navigation turn arrows, distances, and street names onto the HUD, and lets you adjust icon size, resolution, and frame rate. It supports Android 7.0 and above and a Simplified Chinese interface.

Play Steam Games on Android Phones.This app brings SteamOS to Android, letting you run the Steam client on Android devices with Adreno GPUs. It supports running Windows games, installing Linux apps, and signing in to GOG, Epic, and Amazon game stores.

Generating Anki Decks with Python.This Python library generates Anki decks and supports custom note fields, card templates, and CSS styles, along with images and audio. After you modify the content and regenerate the deck, re-importing it will overwrite existing cards with the same names—ideal for batch-creating cards from word lists, question banks, and other data sources.

Making Comics Fit E-Ink Screens.A conversion tool that optimizes comic display for e-ink readers like Kindle and Kobo. It converts images, CBZ, PDF, and other files into e-books, scaling and cropping page margins according to device resolution.

A Faster Math Animation Engine.This project is a 3D math animation engine compatible with the Manim Community API. It uses GPU rendering and is 90x faster than the community edition. It supports AI coding assistants like Claude Code and Codex in writing animations directly, and can run in the browser via WebGPU or export MP4 videos.

Automating Windows Desktop Apps with Python.This is a Python library for Windows GUI automation that can simulate mouse and keyboard actions on windows, dialogs, and controls, and read text data from the UI. It supports two backends: Win32 API and MS UI Automation. It can locate elements by text attributes such as window titles and control names. With just a few lines of code, you can open menus, click buttons, type text, and more. It is suitable for automated testing and repetitive tasks in desktop software.

Writing High-Performance GPU Operators in Python.This is a domain-specific language designed for writing high-performance GPU/CPU/NPU operators. Built on top of the TVM compiler, it lets developers achieve highly optimized implementations of operators such as GEMM and FlashAttention without hand-writing low-level code. It supports multiple hardware backends, including NVIDIA CUDA, AMD ROCm, Apple Metal, and Huawei Ascend, and comes with built-in auto-tuning and debugging visualization tools, plus example operators for large models such as DeepSeek MLA.

Open-Source ESP32 Emulator.This is an ESP32 emulator written in Rust that lets you run unmodified real firmware on your computer or in a browser without a development board. It fully emulates the boot process from the bootloader to FreeRTOS, and can also emulate peripherals such as display, touch, LED, audio, camera, and WiFi, with support for chips including ESP32-S3, C3, C6, and P4.

Minimalist HTTP File-Sharing Server.A command-line tool written in Rust that shares files and directories over HTTP with just a single binary. It supports username/password authentication, file uploads, folder downloads as archives, search by name, QR code access, TLS, and WebDAV. It can also render README.md files like GitHub, offers light and dark themes, and is ideal for scenarios such as temporary file sharing.

NVIDIA's Open-Source Security Sandbox for AI Agents.A secure runtime purpose-built for autonomous AI agents, letting them read and write files, install packages, and call APIs without free rein over your data, keys, or network. Each agent runs in an isolated sandbox, with kernel-level, policy-driven controls over file access, system calls, and network connections. Real credentials are injected only when a request goes to an approved address.

Drawing Technical Concepts as Animated Mechanical Diagrams.An illustration skill for AI coding assistants that turns abstract technical concepts into realistic, movable mechanical devices, rendered as polished isometric interactive illustrations—for example, depicting a cache hit as a lock cylinder that only turns when every pin is aligned. It supports SVG, SVG + WebGL shaders, and 3D, and can export to HTML pages, SVG files, or React components.

Rewrite AI Instructions in Simplified English.This project is based on the Simplified Technical English standard from the aerospace industry. It rewrites ambiguous, verbose English into sentences that won't be misread, then feeds them to AI agents. It checks each sentence for passive voice, complex tenses, multiple instructions in one sentence, long sentences, and marketing adjectives, without losing any facts or conditions from the original.

Let AI Create Design-Forward Diagrams.This is a diagram design skill built for AI coding assistants that generates HTML+SVG files you can open directly in the browser. It supports 45 diagram types, including architecture diagrams, flowcharts, sequence diagrams, Gantt charts, and Sankey diagrams. It can automatically extract brand colors and fonts from websites, redraw existing diagrams from Mermaid, draw.io, and Excalidraw in a consistent style, and export them as PNG or SVG.

Make AI-Designed Interfaces More Tasteful.A front-end design skill for AI coding assistants, extended from Anthropic's official frontend-design skill to help AI avoid cookie-cutter interfaces—such as using Inter everywhere, purple-to-blue gradients, and cards nested inside cards. It provides 24 design commands like polish, audit, and critique, plus 62 deterministic detection rules, and supports Claude Code, Cursor, Codex, Gemini CLI, and other tools.

A macOS Tool That Hangs Screenshots at the Top of Your Screen.A compact macOS screenshot organizer that hangs each screenshot like laundry on a line at the top of your screen. Move your cursor to the menu bar and the line slides down; move away and it automatically tucks itself back, keeping your desktop free from screenshot clutter. Click to copy, press and hold to annotate, drag into an app to send directly, drag to the Trash to delete, and it even supports screen recording and trimming.

A Cross-Platform Xcode Alternative.A cross-platform Xcode alternative written in Swift that lets you build iOS apps with SwiftPM on Linux, Windows, and macOS and deploy them to physical devices. It supports creating projects, building and running apps, signing and installing apps, and managing devices, and it also lets you work with Apple developer services from the command line.

Google's Open-Source AI Phone Automation Tool.This is an AI-powered Android automation tool: just describe a task in natural language, and the AI will automatically operate a phone or emulator to complete cross-app testing workflows and everyday tasks. It includes a built-in MCP server that lets AI coding assistants such as Claude Code and Codex directly control the phone and collect logs and screenshots, and it supports multimodal models including Gemini, Claude, and GPT-4o.

DeepSeek's Open-Source Large Model Inference Framework.This project is the high-performance inference framework used for DeepSeek's large-scale online services, providing core capabilities for offline inference and large-scale deployment. It integrates self-developed optimized operators such as FlashMLA, DeepGEMM, and DeepEP, and supports attention data parallelism, MoE expert parallelism, asynchronous scheduling, deterministic sampling, and structured output. It can run on NVIDIA and Huawei Ascend NPUs.

Google's Open-Source Self-Evolving Agent Framework.This framework enables AI agents to self-improve on unfamiliar tasks without changing model weights. It proposes task-specific modifications—such as adjusting system prompts, memory, skills, sub-agents, and plugins—then runs real trials to compare results, and finally captures the lessons learned for future tasks.

Animate Any 3D Character with One Sentence.This project can generate motions for 3D models with different skeletal structures. Given a rigged 3D model and a text description, it can generate the corresponding motion in real time without retraining for each skeleton. It works with a variety of skeletal structures, including humanoids, quadrupeds, birds, marine creatures, insects, and mechanical rigs.

Open-Source Local AI Voice Studio.This is a fully local voice cloning and dubbing tool that serves as an open-source alternative to ElevenLabs and supports 600+ languages. It integrates multiple speech synthesis and recognition engines such as CosyVoice, IndexTTS, GPT-SoVITS, and WhisperX, with features including voice design, video dubbing, vocal separation, transcription, and audiobook production.

Building Open-Source Alternatives to Commercial Software from Scratch.This is an open-source project initiated by the Puter team, aiming to rebuild free, open-source alternatives to commercial software from scratch. It currently offers over 50 apps, including alternatives to Excel, PowerPoint, Photoshop, AutoCAD, Zoom, and Slack, as well as small games like chess and Sudoku—most of which can be used directly in the browser.

A Cyberpunk-Style Periodic Table.An interactive periodic table of chemical elements that runs in the browser with a dark cyberpunk aesthetic. It covers atomic number, electron configuration, isotopes, and valence data for all 118 elements. You can switch heatmaps based on properties such as atomic radius, electronegativity, and melting point, click any element to drag and rotate a 3D atomic model, explore its electron shell structure, and use a bilingual Chinese/English interface.

Unlocking the Compute Power of RTX 30 Series GPUs.This is a Linux kernel module that unlocks the FP16/BF16 instruction issue rate of NVIDIA RTX 30 series consumer GPUs to the level of their professional counterparts, significantly boosting matrix operation throughput. For example, the BF16 compute performance of the RTX 3090 Ti can increase from 87.5 to 157.4 TFLOPS, and large model prefill speed improves by 14%–36%. It performs the unlock before the official NVIDIA driver loads, requiring no driver modifications and no need to reapply patches after each driver update.

Open-Source Mobile Robot Navigation Framework.This is an autonomous robot navigation framework that provides perception, localization, path planning, motion control, and obstacle avoidance, with behavior trees orchestrating modular services. Beyond basic navigation, it also supports waypoint cruising, target following, and full-coverage navigation tasks.

A Minimalist Local Image Compression Tool.This is a local processing tool for images and animated images. It supports batch compression, drag-to-compare quality, format conversion, resizing, and watermarking. It automatically compares candidate outputs such as JPEG, WebP, and PNG, then selects the smaller file. It offers desktop apps for Windows, macOS, and Linux, and can also be used in the browser or via Docker. All images are processed locally on your machine. The desktop version also supports folder monitoring, automatic clipboard optimization, and MCP integration.

A Cost-Effective Guide to Life.A practical guide to trading less money, time, and energy for more years of life and greater financial security. It covers longevity and disease prevention, first aid, saving and money management, legal red lines, unemployment assistance, medical and social insurance, dating, marriage, and childbearing, and compliance for entrepreneurs.