Trending AI projects

ms‑365‑mcp‑server is a Node.js MCP server that wraps Microsoft 365 Graph APIs as a large set of declarative “tools” for LLM assistants. It handles MSAL authentication, lets you filter the tool surface and OAuth scopes, supports read‑only mode, and can output results in JSON or the token‑saving TOON format. Designed for Claude Desktop, Open WebUI, and similar AI‑agent platforms.

Spark is an open‑source, MIT‑licensed library that adds real‑time 3‑D Gaussian‑splatting rendering to THREE.js. It works on almost any WebGL2‑capable device, supports many splat file formats, and provides dynamic editing, multi‑view rendering, and a shader‑graph system. Install via CDN or npm, or build locally (Rust → WASM). Created by World Labs.

A curated list of open-source libraries for deploying, monitoring, versioning, scaling, and securing production machine learning systems, organized by category with monthly updates and a search tool.

A community-curated directory of Agent Skills — simple text files that teach AI assistants how to do specific tasks on demand. It organizes official skills from major AI companies (Anthropic, OpenAI, Google, etc.) and community-contributed skills, with tools for finding and managing them. Skills load in three stages (browse, load, use) and work across many AI agents like Claude Code, Codex, and GitHub Copilot.

memU is a lightweight, agent‑driven personal memory system that records LLM session logs, turns useful interactions into reusable Markdown “skills”, and makes those skills searchable across sessions, agents, and devices. It works via small host‑adapter binaries (e.g., memu‑codex, memu‑claude‑code) that hook into agents’ session logs and instruction files. The core memory logic is ~500 lines of Python, storing embeddings in SQLite, Postgres, or an in‑memory store. Installation is a one‑message setup with an API key from memU Cloud (or self‑hosted). Once installed, agents automatically retrieve relevant skills before answering, and developers can feed custom conversations into memU via `memu memorize`. Licensed under Apache‑2.0.

A Claude Code plugin that delegates heavy or repetitive development tasks to a cheaper Gemini-based agent, allowing Claude to focus on high-level judgment and verification while reducing overall costs.

Vosk is an offline open-source speech recognition toolkit that provides continuous transcription for over 20 languages across devices ranging from Raspberry Pi to large clusters.

LLPlayer is a Windows‑only C# media player that adds AI‑driven subtitle generation (Whisper ASR), real‑time translation (via cloud or local LLMs), OCR for bitmap subtitles, dual‑subtitle display, word lookup, and online video support. It targets language learners and is open‑source under GPL‑3.0.

An open-source autonomous driving framework providing a production-ready software stack for the commercial deployment of autonomous vehicles.

TheBigPromptLibrary is an open, educational collection of real system prompts, custom instructions, jailbreak prompts, and protection prompts for major AI chatbots, plus articles and tools about extracting and securing them. It explicitly allows only lawful, educational use.

This Obsidian plugin exposes the vault via a secure, authenticated REST API and a built-in MCP server, giving scripts, browser extensions, and AI agents full CRUD, search, targeted patching, event streaming, and command execution while blocking access to the config directory.

ShardX is an open-source anti-detect browser launcher that runs isolated, fingerprint-spoofed Chromium profiles to help users manage multiple identities and perform web scraping without being detected.

An interactive visualization tool that runs a live GPT-2 model in the browser to help users learn how Transformer-based models predict text.

A Claude Code skill that enables AI to generate professional consulting-style presentation decks using a strict design rulebook, a 62-part HTML library, and automated rule checking.

RoboCasa365 is an open‑source simulation suite for training and benchmarking generalist robot manipulation policies in kitchen environments. It offers 365 LLM‑generated tasks, thousands of scenes and objects, extensive demonstration data with sub‑task labels, Gym wrappers, and a public leaderboard for comparing algorithms.

An MCP server that enables AI assistants to interact with Obsidian vaults by reading, searching, and editing notes via the Local REST API plugin.

LinguaGacha is a desktop app that uses LLM APIs (OpenAI, DeepSeek, etc.) to batch‑translate novels, subtitles, and game scripts across 16 languages, preserving formatting and offering term‑extraction and automatic proofreading.

A scalable social media simulator that uses LLM agents to mimic up to one million users on platforms like Twitter and Reddit to study social phenomena.

A real-time audio translation tool for Windows 11 that integrates Windows Live Captions with LLM and traditional translation APIs to provide translated subtitles via a transparent overlay.

Documentation for Figma’s MCP server – an HTTP service that lets AI coding assistants read design data (components, variables, layout) from Figma, generate placeholder images, and (in beta) write back to the canvas. Includes setup for VS Code, Cursor, Claude Code, Gemini CLI, rate‑limit info, and best‑practice tips for prompting agents to turn frames into code.

An AI agent framework designed to automate mathematical modeling competitions, from problem analysis and coding to generating professional submission-ready papers.

MJWarp is a GPU-accelerated physics simulator for robotics research that provides high-throughput simulation and batch rendering on NVIDIA hardware.

Adds custom nodes to ComfyUI that let you chain MiniMax‑H3 video‑plus‑audio generations. The previous clip is passed as a latent (no decode‑encode), preserving exact pixel data and continuing the audio waveform. Includes Save/Load Latent, Trim, Chain, and optional drift‑reduction nodes, enabling seamless multi‑clip videos with consistent motion and sound.

An open-source framework for quadruped robots that handles gait coordination and servo control, enabling researchers and makers to build AI-powered robotic pets.

A Chinese translation of the book 'Build a Large Language Model (From Scratch)', providing a technical guide and code examples for understanding and implementing LLM architectures.

MinhNguyenDS/AI-pdf-books is a curated, folder-organized collection of AI/ML/LLM/data-science PDFs from major publishers, with clear contribution guidelines and a legal disclaimer stating the materials are for educational use only and remain under their original copyrights.

A repo of MCP servers that let AI assistants read, analyze, and modify Cloudflare account configurations via natural language. Includes one recommended broad-access API server (in a separate repo) plus domain-specific servers for documentation, containers, browser rendering, monitoring, security, and more. Each server exposes a stateless HTTP handler at /mcp; the old /sse endpoint returns a 410 error. Connect via any remote-capable MCP client or Cloudflare AI Playground. Some features require a paid Workers plan.

Loop Engineering for VLAnything is an end‑to‑end Python toolkit that lets you record, merge, audit, enrich, and iteratively improve multimodal robot‑learning datasets (RGB, RGB‑D, and future visuo‑tactile) for vision‑language‑action models. It provides a conservative, side‑car‑first quality audit, plug‑in semantic evaluators, language‑annotation pipelines, and a closed‑loop feedback loop for policy improvement, all compatible with the LeRobot data format and publishable to Hugging Face.

ripwire is a one-binary tool that gives AI coding agents a fast, cheap, and honest map of a codebase, cutting token usage and preventing bugs during automated edits.

A companion codebase for the O'Reilly book *AI Systems Performance Engineering*, offering profiling scripts, CUDA/Triton kernels, distributed‑training configs, and a 200‑item performance checklist to help engineers optimise GPU‑based training and inference.

A deep learning library built on PyTorch that provides high-level components for rapid productivity and low-level components for research flexibility.

Nx is a smart monorepo tool for TypeScript and polyglot codebases that uses incremental builds and AI-powered CI self-healing to accelerate development cycles.

Memmy is a local memory and agent runtime that provides persistent context across different AI coding assistants, allowing them to share project history and preferences.

An AI-powered tool that uses Florence-2 and LaMA models to detect and seamlessly remove watermarks from images and AI-generated videos.

learn-workbuddy is an open-source teaching blueprint that rebuilds the architecture of the WorkBuddy desktop AI assistant in 24 hands-on Python lessons, each adding one mechanism (agent loop, sidecar, memory, audit sandbox, etc.). It solves the engineering gap between CLI agents and production desktop agents — long-lived processes, context limits, safety, and multi-provider support — with fully offline demos and a minimal standard-library reference implementation.

AIOS is an open‑source “operating system” for LLM‑based agents. The kernel (this repo) manages LLM instances, memory, storage, and tool sandboxing, exposing a syscall‑style API via the Cerebrum SDK. It supports many model back‑ends (Ollama, vLLM, HuggingFace, Deepseek), integrates external tools, and offers several deployment modes—from local single‑machine setups to remote, personal, and virtualized kernels. Installation is via a Python 3.10/3.11 environment, configuration is done in `config.yaml`, and agents can be run through a web UI or a natural‑language terminal. The project is research‑grade (multiple conference papers) and actively maintained.

A self-hosted job search assistant that uses AI to automate job discovery, review resumes, and match candidates to job listings while keeping data private.

OpenWrite is a DeepSeek Harness plugin that adds a complete AI‑assisted novel‑writing workflow—planning, drafting, multi‑model review, and export—via a web UI and slash commands, storing all assets locally.

JADX‑AI‑MCP is a Java plugin for the JADX decompiler plus a Python MCP server that lets large language models (e.g., Claude) interact with a decompiled Android APK in real‑time. Users can ask the AI to explain code, find vulnerabilities, rename obfuscated symbols, fetch resources, and even debug, all from within the JADX GUI.

Mural is a native iPhone/Android app that uses your own OpenAI API key to run AI‑driven spoken conversations for language learning. It stores progress locally, adapts difficulty from your replies, and supports 12 languages out of the box.

An AI-powered interview platform that provides resume analysis, RAG-based knowledge base management, and real-time voice/text mock interviews with automated evaluation.

FlowGram is an open‑source visual workflow framework (bytedance/flowgram.ai) for building AI‑agent pipelines. It offers a web SDK and a native desktop app with drag‑and‑drop canvases, built‑in LLM, HTTP, and code nodes, scoped variables, form generation, debugging, and versioned publishing. Quick start via `npx @flowgram.ai/create-app@latest` and extensive docs at flowgram.ai.

Meeting Recorder is a Rust‑based Linux app that records microphone and system audio, transcribes locally with Whisper, provides speaker diarisation, optional AI‑generated chapters, and integrates tightly with the Omarchy desktop environment—all while keeping data on‑device.

A private, local-first voice and text AI assistant that maintains a persistent memory of conversations and integrates with local tools to act as a personal companion.

A multi-agent framework for building socio-collaborative AI companions that prevents persona collapse and social sycophancy through bi-level optimization of individual speakers and a coordinating director.

ProtoMotions 3 is an Apache‑2.0‑licensed, GPU‑accelerated framework for large‑scale motion learning, retargeting, and sim‑to‑real deployment of humanoid robots. It supports multi‑GPU training on datasets such as AMASS and BONES, one‑command retargeting via PyRoki, cross‑simulator testing (Isaac Gym, Newton, MuJoCo), and export of ONNX policies for zero‑shot deployment on real hardware (e.g., Unitree G1).

Airbyte is an open-source data movement platform that syncs data from hundreds of sources to warehouses or AI agents, providing an SDK to turn these connectors into LLM-ready tools.

HoloMotion is an end-to-end framework for humanoid robot control that enables real-time motion tracking and command following by converting diverse human motion data into actionable robot policies.