Gaurav Gosain

xy. Ultra-fast and customizable Python charts

Modus. MODUS: Decoder-only Any-to-Any Modeling of Diverse Modalities [ICML 2026]

IRIS. A shell auto-completion tool for your terminal

neru. Navigate your entire screen without touching the mouse.

kimi-k3-mlx. MLX port of moonshotai/Kimi-K3 (2.78T multimodal MoE): streaming converter, REAP expert pruning, and per-language expert-overlap analysis

vicinae. A focused launcher for your desktop - native, fast, extensible

gigatoken. Language model tokenization at GB/s

PGSimCity. An explorable 3D city that shows how Postgres actually works

xls. XLS: Accelerated HW Synthesis

tinycast. Tinycast — a tiny, fully native macOS launcher, hotkeys, and clipboard history

uncurses. A terminal toolkit library for building UIs

scriptc. TypeScript-to-Native Compiler

firefox-wasm. 🦊 Firefox in WebAssembly

wisp-protocol. Wisp is a low-overhead, easy to implement protocol for proxying multiple TCP/UDP sockets over a single websocket.

reims-vgpu. reims-vgpu is an experimental virtual GPU for macOS guests

colibri. Run GLM-5.2 (744B MoE) on a 25GB-RAM consumer machine — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦

qemu. Official QEMU mirror. Please see https://www.qemu.org/contribute/ for how to submit changes to QEMU. Pull Requests are disabled. Please only use release tarballs from the QEMU website.

qemu. qemu with iOS host support

lexical. Lexical is an extensible text editor framework that provides excellent reliability, accessibility and performance.

iDescriptor. A free, open-source, and cross-platform iDevice management tool

shadscan. Deterministic UI audits for shadcn apps, built for your terminal, your CI, and your AI agent.

bklit-ui. Open-source UI & Charts library

VimItAll. System-wide Vim mode for macOS via Accessibility API and keyboard fallback

muslnet. TinyGo support for networking on Linux

fp-go. Functional programming library for Go 1.24+, inspired by fp-ts. Uses generic type aliases for a clean, composable API. Provides Option, Either, Result, IO, IOResult, Reader, and ReaderIOResult monads, plus optics (Lens, Prism, Traversal) for immutable data manipulation. Supports Functor, Applicative, and Monad abstractions with do-notation-style

calendar. Calendar in your terminal

starlight. 🌟 Build beautiful, accessible, high-performance documentation websites with Astro

satteri. High-performance Markdown and MDX processing for the JavaScript ecosystem

nimbus. Docs for humans and agents, built on Astro

headline. CLI tool to make headers from simple config files

kumo. Cloudflare's component library for building modern web applications.

img2threejs. Rebuild the object in a reference image as a code-only, procedural, quality-gated, animation-ready Three.js model. Token-efficient image-to-3D.

dotlottie-rs. High-performance Lottie & dotLottie player in Rust, with bindings for Android, iOS, Web (WASM), and C/C++.

dotlottie-web. High-performance Lottie & dotLottie web player powered by Rust + WASM, with React, Vue, Svelte, Solid & Web Components.

wazy. Fastest WebAssembly runtime for Go. Supports Component Model & Async. Zero dependencies, no CGO, pure Go.

davinci-resolve-mcp. MCP server integration for DaVinci Resolve Studio

elements. Write once in React. Render emails, web pages, and PDFs from the same component tree.

webterm. An embeddable browser terminal built on xterm.js, with kitty graphics, a clipboard that works on insecure origins, and reference-checked unicode widths.

lenis. Smooth scroll as it should be

Ontology-Playground. Free, open-source web app for learning about ontologies and Microsoft Fabric IQ. Explore a catalogue of pre-built ontologies, design your own visually, export as RDF/XML, and share interactive diagrams. Zero backend, fully static.

tuitest. Headless testing harness for terminal programs: real PTY, VT emulator, deterministic waits, and golden screen snapshots

vtgl. WebGL2 glyph-atlas terminal renderer driven by a pluggable read-only VT source, with a Canvas2D fallback

cuelume. Cuelume is a curated sound palette, not an audio engine. It gives buttons, links, toggles, and completed actions clear feedback without asking developers to design sounds themselves. Add an attribute, call bind(), done.

zaly. Hackable terminal coding agent

frame. FFmpeg GUI

hallmark. Anti-AI-slop design skill for Claude Code, Cursor, and Codex.

grok-build. SpaceXAI's coding agent harness and TUI. Fullscreen, mouse interactive, extensible.

cadvisor. Analyzes resource usage and performance characteristics of running containers.

nemotron-voice-agent. Reference implementation of an end-to-end voice agent built using the NVIDIA Nemotron models

upx. UPX - the Ultimate Packer for eXecutables

scroll-world. A skill that turn any brand into a scrollable 3D world

lingbot-vision. Self-supervised learning for spatial perception

MOSS-VL. MOSS-VL is the core multimodal model series within the OpenMOSS ecosystem, dedicated to visual understanding.

proxypose. Official repository for the paper "ProxyPose: 6-DoF Pose Tracking via Video-to-Video Translation"

qmd. mini cli search engine for your docs, knowledge bases, meeting notes, whatever. Tracking current sota approaches while being all local

qwopus-dflash. DFlash speculative decoding for Qwopus3.5-9B-Coder on upstream llama.cpp

oyehoye. AMD Hackathon

skills. Skills for Design Engineers.

spacewasm. A flight-compliant WebAssembly interpreter.

perry. A native TypeScript compiler written in Rust. Compiles TypeScript directly to executables using SWC and LLVM.

MOSS-Transcribe-Diarize. MOSS-Transcribe-Diarize 0.9B is an open-source SOTA end-to-end audio understanding model for long-form multi-speaker transcription, diarization, timestamps, and acoustic event awareness.

RF-DETR-CPP. C++

BuildingGeneratorThreeJS. TypeScript

agentation. The visual feedback tool for agents.

impeccable. The design language that makes your AI harness better at design.

riddle. The diary of Tom Riddle for the reMarkable Paper Pro — write with your pen, the page drinks your ink and answers in a flowing hand

pocket-tts. A TTS that fits in your CPU (and pocket)

shepherd. A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent and environments in a copy-on-write fork ~5x faster than docker commit, with ~95% KV-cache reuse on replay. Framework built for meta-agents to supervise, optimize, and train other agents

LEANN. [MLsys2026]: RAG on Everything with LEANN. Enjoy 97% storage savings while running a fast, accurate, and 100% private RAG application on your personal device.

zero. The coding agent that answers to you, your model, your machine, your rules.

GenRecon. Python

react-native-client. Native React Native HTTP client for resumable background file downloads.

astryx. An open source design system that's fully customizable and agent ready

oasdiff. OpenAPI Diff and Breaking Changes

vacuum. vacuum is the worlds fastest and most versatile OpenAPI, AsyncAPI & JSON Schema linter, docs generator and toolkit. It tears through API specs at light speed. 100% compatible with Spectral rulesets, and OpenAPI 3.0, 3.1 and 3.2

regexplain. A terminal UI for explaining and visualizing regular expressions, kindof like regex101.

fast. Internet speed test in your terminal

Extend3D. Jupyter Notebook

FluidVoice. Fastest and only macOS Dictation app with on-device STT and custom trained AI enhancement model. A local Wispr Flow alternative. ⭐ helps a ton :) Windows & iOS waitlist open. Linux soon.

VisualClaw. Official Implementation of VisualClaw: A Real-Time, Personalized Agent for the Physical World

lore. Lore is a next-generation, open source version control system

JetSpec. JetSpec: Breaking the Scaling Ceiling of Speculative Decoding with Causal Parallel Tree Drafting

terrain-diffusion. Procedural generation with diffusion models (SIGGRAPH '26)

free-splatter.cpp. Turn images into 3D scenes - GGML port of FreeSplatter: Pose-free Gaussian Splatting for Sparse-view 3D Reconstruction

ttop. A tmux-aware process tree with top-style CPU/memory views

threejs-water. A real-time water simulation with Three.js, featuring raytraced reflections, refractions, caustics, and interactive objects.

Tater. Tater is a local-first AI platform with built-in LLMs, vision, voice, memory, automation, integrations, and voice satellite support.

jose. JWA, JWS, JWE, JWT, JWK, JWKS for Node.js, Browser, Cloudflare Workers, Deno, Bun, and other Web-interoperable runtimes

video-search-and-summarization. NVIDIA AI Blueprint for video search and summarization (VSS) is a GPU-accelerated reference architecture for building video analytics agents with real-time verified alerts, visual Q&A, and automated reporting. The VSS Blueprint uses vision language models (VLMs) such as NVIDIA Cosmos, LLMs such as NVIDIA Nemotron, RAG, and NVIDIA NIMs.

Qwen-AgentWorld. Qwen-AgentWorld: Language World Models for General Agents

bun-sqlgen. Types generator for your Bun.sql queries

flue. The sandbox agent framework.

OpenCut. The open-source CapCut alternative

codebase-memory-mcp. High-performance code intelligence MCP server. Indexes codebases into a persistent knowledge graph — average repo in milliseconds. 158 languages, sub-ms queries, 99% fewer tokens. Single static binary, zero dependencies.

ArtiFixer. Python

ScaRF-SLAM. ScaRF-SLAM: Scale-Consistent Reconstruction with Feed-Forward Models and Classical Visual SLAM

iptop. htop for your network; a beautiful, real-time terminal IP traffic monitor

LocalVQE. Lean neural real-time acoustic echo cancellation with soft delay estimation - GGML and PyTorch inference

hiring-agent. AI agent to evaluate and score resumes.

ponytail. Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.

webgl-water. WebGL Water Demo

R3. Official implementation of R^3: 3D Reconstruction via Relative Regression.

gsap-skills. Official AI skills for GSAP. These skills teach AI coding agents how to correctly use GSAP (GreenSock Animation Platform), including best practices, common animation patterns, and plugin usage.

ShojiWM. The most customizable Wayland compositor with TypeScript(tsx).

depth-anything.cpp. A from-scratch C++17/ggml port of Depth Anything 3 (ByteDance)

slime. slime is an LLM post-training framework for RL Scaling.

kernel-images. Browsers-as-a-service for automations and web agents

hypeman. A multi-hypervisor VM runtime for OCI images, supporting Cloud Hypervisor, Firecracker, QEMU, and Apple Virtualization.framework.

slang-splat. gaussian splatting implemented using slangpy

turbograph. Fast, local, hackable graph RAG engine in Go: hybrid dense+lexical retrieval, PageRank over a chunk+entity graph, community summaries for global queries, multimodal, single binary, local Ollama.

clin-rs. Feature-packed TUI note management app inspired by Obsidian

PERSIST. Codebase for "Beyond Pixel Histories: World Models with Persistent 3D State" (ICML 26)

privacy-filter.cpp. OpenAI's privacy filter NER model architecture implemented in a minimal C++/GGML runtime

lore-go. Go SDK for Lore, a next-generation, open source version control system

cc-oyehoye. ClaudeCode hook that plays "Oye Hoye Haye Haye" every time Claude finishes responding. Courtesy of Chahat Fateh Ali Khan.

PaddleOCR. Turn any PDF or image document into structured data for your AI. A powerful, lightweight OCR toolkit that bridges the gap between images/PDFs and LLMs. Supports 100+ languages.

OpenEnv. An interface library for RL post training with environments.

firnflow. The cost efficiency of S3 with the speed of local RAM. A multi-tenant vector and full-text search engine featuring a tiered RAM -> NVMe -> S3 architecture for microsecond latency on top of object storage

open-design. 🎨 The open-source Claude Design alternative. 🖥️ Local-first desktop app. 🖼️ Your coding agent becomes the design engine: prototypes, landing pages, dashboards, slides, images & video — real files, HTML/PDF/PPTX/MP4 export. 🤖 Claude Code / Codex / Cursor / Gemini / OpenCode / Qwen & 20+ CLIs via BYOK.

ku. A fast, keyboard-driven Kubernetes TUI. Browse any resource, edit objects, follow logs, and shell into pods.

meshflow. Repository for the CVPR 2026 paper MeshFlow Efficient Artistic Mesh Generation via MeshVAE and Flow-based Diffusion Transformer by Weiyu Li, Antoine Toisoul, Tom Monnier, Roman Shapovalov, Rakesh Ranjan, Ping Tan and Andrea Vedaldi.

small-talk. An AI-to-AI robot podcast hosted by Reachy Mini robots — live LLM+TTS generated shows, 3D twins, a radio station, and a physical-robot companion

llama-modal-serve. serving various LLMs via modal.com

UniSHARP. Official implementation of UniSHARP: Universal Sharp Monocular View Synthesis

Telegram-iOS. Telegram-iOS

LatentSpatialMemory. Latent Spatial Memory for Video World Models

clawpatrol. Security firewall for agents

onnxruntime. ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator

EcoDepth. [CVPR'2024] Official implementation of the paper "ECoDepth: Effective Conditioning of Diffusion Models for Monocular Depth Estimation"

lottie. Generate production-ready Lottie animations with Claude Code or Codex

LAMP. [CVPR 26'] Code for the LAMP research paper.

kyverno. Unified Policy as Code

jamstation. A Playstation One (PS1) emulator fully written in Jam

gg-svg. SVG export backend for gg's recording system

gg-pdf. PDF export backend for gg's recording system

gg. Enterprise-grade 2D graphics library for Go. Pure Go, zero CGO. Part of GoGPU ecosystem.

gogpu. Pure Go GPU Application Framework — windowing, input, lifecycle, platform abstraction. Part of the GoGPU ecosystem.

xet-core. xet client tech, used in huggingface_hub

turbovec. A vector index built on TurboQuant, written in Rust with Python bindings

beeket. Run LLM with Yzma

I. I -> C transpiler

MisoTTS. Miso TTS is an 8 billion, highly emotive text-to-speech model

ideogram4. Ideogram 4: Open image model at the forefront of design

genai. The opinionated high performance professional-grade AI package for Go

gh-stack. GitHub Stacked PRs

draco.js. A small, drop-in, pure-JavaScript Draco mesh loader for three.js.

jjdag. A TUI to manipulate the Jujutsu DAG. LLM slop feature completion for personal use; author can ask if they want the features. Don't wanna PR DDoS 'em

bigset. What if you had all the data in the world?

gooey. Gooey is a hybrid immediate/retained mode UI framework designed for building fast, GPU-rendered applications on macOS/Metal, WebAssembly/WebGPU, and Wayland/Vulkan

tuxedo. A fast, keyboard-driven terminal UI for todo.txt.

taste-skill. Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop

coreutils. Coreutils for Windows: Installer & Packaging

pgxcli. Postgres CLI with autocompletion and syntax highlighting

sortie. AI harness that you actually pilot

ScrollWheel. Minimalist RP2350 magnetic sensor scroll wheel toy project

mxc. Policy-driven, layered isolation and containment

BaseLayer. Browser hosting control plane and host runtime for running Chromium sessions

libreyolo. LibreYOLO is a MIT licensed open source computer vision library

ml-pico. JavaScript

parakeet.cpp. Parakeet implementation in C++ with ggml

odysseus. Self-hosted AI workspace.

walls. All of my wallpapers in one repo.

dvlt. Official implementation of Déjà View: Looping Transformers for Multi-View 3D Reconstruction

REST3D. From a single casual image to a visually consistent and physically stable interactive 3D scene.

vim-dogrun. :dog: A dark Neovim / Vim colorscheme for the GUI and 256 / true-color terminals.

liteparse. A fast, helpful, and open-source document parser

qr.nvim. QR code generator for neovim

VoxInput. 🎤 A voice assistant that let's you control any Linux desktop and transcribe any audio.

LocalAI-examples. LocalAI examples

SAM3DBody-cpp. Real-time 3D full-body reconstruction from a single camera, Multiperson BVH output, Pure C++ runtime, ONNX + ggml, 70-joint skeleton with hands.

SysMocap. A real-time motion capture system for 3D virtual character animating.

TriSplat. TriSplat: Simulation-Ready Feed-Forward 3D Scene Reconstruction

ghui. A GitHub TUI

Eagle. Eagle: Frontier Vision-Language Models with Data-Centric Strategies

FastVideo. A unified inference and post-training framework for accelerated video generation.

openWakeWord. An open-source audio wake word (or phrase) detection framework with a focus on performance and simplicity.

silero-vad. Silero VAD: pre-trained enterprise-grade Voice Activity Detector

quien. A better whois and domain intelligence toolkit

Bonsai-Image-Demo. Generate images locally

aholo-viewer. A high performance 3DGS renderer

ppf-contact-solver. A contact solver for physics-based simulations involving 👚 shells, 🪵 solids, 🪢 rods, 🧱 rigid bodies and ⏳ sand.

vgg-ttt. [CVPR'26] Official code for the paper "VGG-T³: Offline Feed-Forward 3D Reconstruction at Scale"

bumblebee. Read-only developer endpoint scanner for on-disk package, extension, and developer-tool metadata, built to check exposure to known software supply-chain compromises.

iree. A retargetable MLIR-based machine learning compiler and runtime toolkit.

Pixal3D. [SIGGRAPH 2026] Pixal3D: Pixel-Aligned 3D Generation from Images

ml-lito. [ICLR 2026] LiTo: Surface Light Field Tokenization

android-titanium-browser. Secure open-source Android browser with support for extensions

wakaru. 🔪📦 Javascript decompiler for modern frontend

pgvector. Open-source vector similarity search for Postgres

RADIO. Official repository for "AM-RADIO: Reduce All Domains Into One"

alien-signals. 👾 The lightest signal library

brownies. 🍫 Tastier cookies, local, session, and db storage in a tiny package. Includes subscribe() events for changes.

supertonic-py. Lightning-Fast, On-Device TTS — running natively via ONNX.

PhotoGIMP. A Patch for GIMP 3+ for Photoshop Users

Lance. A 3B-active-parameter native unified multimodal model for image and video understanding, generation, and editing.

ai-hub-models. Qualcomm® AI Hub Models is our collection of state-of-the-art machine learning models optimized for performance (latency, memory etc.) and ready to deploy on Qualcomm® devices.

hexagon-mlir. Hexagon-MLIR is a compiler toolchain for compiling and executing AI kernels and models on Qualcomm Hexagon Neural Processing Units (NPUs).

cpullvm-toolchain. LLVM

fastrpc. FastRPC is Qualcomm's userspace library that facilitates efficient remote procedure calls between the CPU and DSP for high-performance computing.

cli. One CLI for all your chats