Trending AI projects

A collection of agent skills and plugins for Google Stitch that enables AI coding agents to automate design generation, design system management, and conversion of designs to frontend code.

Easy-Vibe is a comprehensive, multilingual learning guide that teaches users how to build software products from scratch using AI-assisted coding workflows.

LazyMind is a local‑first AI‑Skill Runtime that turns knowledge bases, reusable Skills, and defined workflows into observable, editable, and versioned AI‑driven tasks. It offers a desktop app (macOS/Windows) and an enterprise‑grade container deployment, supports OCR, hybrid retrieval, step‑by‑step workflow execution, and integrates with agents like Codex and DeepSeek Harness.

A native Zig-based inference server for Apple Silicon that supports MLX and GGUF models with OpenAI, Anthropic, and Ollama API compatibility.

An open-source framework that enables AI agents to automate video production by using reusable templates to handle photo placement, copy, and rendering based on simple user prompts.

A marketplace of specialized plugins for AI coding assistants that adds expert-level security auditing, vulnerability research, and advanced development workflows to agents like Claude Code.

A comprehensive toolkit for generating, syncing, and managing professional .drawio architecture diagrams from natural language, source code, and infrastructure configurations.

Full Chain Sentinel System is an authorized penetration-testing agent that combines an AI assistant with per-project Kali containers, a bilingual web console, and tool integrations for asset discovery, web research, and mobile testing—built for security teams needing a controlled, audited, and multi-platform offensive security workflow.

An open-source backend platform for agentic coding that provides AI agents with database, auth, storage, and compute to build full-stack apps autonomously.

A skill for AI agents that turns a one‑sentence brief into a full, award‑quality website – handling concept, design system, code generation, automated QA and optional deployment.

Nano-vLLM is a lightweight, from-scratch implementation of a vLLM-style inference engine designed for fast offline inference with a highly readable codebase.

GRAVA is a vision-language-action framework for autonomous driving that connects visual scene understanding to executable driving trajectories through grounded reasoning.

HiP3R is a framework for 3D reconstruction that processes long video sequences by aligning overlapping chunks and refining camera poses to create consistent spatial models.

An open-source computer use agent framework that operates a real GUI via mouse, keyboard, and screen to complete natural-language tasks across desktop and web applications.

NarratoAI is an open‑source desktop/web application that automates video narration: it uses LLMs to write scripts, cuts video highlights, generates voice‑overs with various TTS engines (including Apple‑Silicon MLX models), creates subtitles via ASR, and can add AI‑generated music. Install via a pre‑built Windows/macOS bundle, Docker, or from source with Python 3.12+. The UI runs on Streamlit at http://127.0.0.1:8501. Licensed under an open‑source license.

A collection of 15 modular skills for Claude Code that enables users to generate professional motion graphics and animated videos entirely through code.

A local 12B parameter model and desktop application that rewrites AI-generated drafts to sound human while preserving factual accuracy and formatting.

open‑computer‑use (aka `ocu`) is an open‑source, cross‑platform Computer‑Use service that exposes macOS, Linux and Windows UI automation via a standard MCP JSON‑RPC server. It lets AI agents (Codex, Claude Code, Gemini CLI, etc.) run actions like listing apps, pressing keys, or querying UI elements, and includes a JavaScript REPL, skill packages for popular agents, permission helpers, and an optional macOS cursor‑motion subsystem.

Knowhere is an open‑source backend that parses PDFs, slides, Word docs, images, etc., into a hierarchy‑native memory graph. It offers dual‑track (text + vision) parsing, agent‑neutral retrieval APIs, and citation‑ready results, enabling agents to navigate and cite complex documents efficiently.

NeonPlan 3D is a Home Assistant integration that lets you draw a floor‑plan and interact with a neon‑styled 3‑D view of your house, controlling lights, blinds, doors, cameras and more, with optional paid furniture packs and Pro add‑ons.

RD‑Agent is an open‑source Microsoft framework that lets LLM‑driven agents automatically propose ideas (R‑agent) and implement them (D‑agent) across data‑centric R&D tasks such as quantitative‑finance factor modelling, Kaggle‑style data‑science pipelines, and research‑paper code generation. It tops the MLE‑bench benchmark, supports any LiteLLM‑compatible model, and provides Docker‑based demos and a web UI for interactive tracing.

A self-hosted AI stock monitoring platform that combines portfolio management with a multi-agent research team for deep analysis of A-shares, Hong Kong, and U.S. stocks.

SkillOpt is a Microsoft‑open‑source framework that optimises an LLM‑agent’s skill prompt as a trainable document using a deep‑learning‑style loop, delivering measurable performance gains without altering the model’s weights.

An unofficial Python client/CLI for Google’s Gemini Notebook (formerly Notebook LM). It lets you program‑matically create notebooks, add many source types, ask Gemini‑powered questions, generate audio/video/slides/quiz/flashcards/mind‑maps, and batch‑download the results. Designed for research automation, LLM‑agent integration (Claude Code, Codex, etc.), and head‑less scripting via a Python API, MCP server, or REST server. Uses undocumented Google APIs, so it may break and is rate‑limited.

This is a complete DIY kit for building a walking robot duck — replica of the open-source Microduck, using Feite HD1910 servos. It ships pre-trained motion models (no AI training needed), a Chinese control console, custom circuit boards, 3D-print files, and a step-by-step build guide. The duck can walk, sit, kick, roll, and even balance on one leg (author's self-trained breakthrough). Free for personal, non-commercial use; commercial use needs written permission.

DeepEval is an open‑source Python framework that lets you write pytest‑style tests for any LLM‑based app—chatbots, agents, RAG pipelines, or multimodal generators. It ships with dozens of research‑backed metrics (e.g., G‑Eval, JevEval, task‑completion, hallucination, bias) that can run on any model locally, and you can add custom metrics. The library integrates with popular toolkits (LangChain, OpenAI, Anthropic, LlamaIndex, etc.) to automatically trace model calls, enabling both end‑to‑end and step‑level evaluation. Tests run from the CLI, fit into CI/CD, and can be linked to the commercial Confident AI platform for production‑time monitoring. Ideal for developers and teams who need quantitative quality gates for LLM applications.

Handwired converts regular JavaScript code into a differentiable program made only of matrix multiplication and ReLU, enabling GPU‑accelerated execution and gradient‑based experiments without any training.

A bridge that connects AI agents running on a local machine to instant messaging platforms, allowing remote control of local agents via chat apps.

A native desktop application for the DeepSeek Harness agent platform that simplifies installation, runtime management, and plugin handling while enabling remote mobile access.

A DeepSeek Harness plugin that enables users and AI to search, send, and learn memes, supporting both manual selection and automated AI-driven meme battles across web and IM platforms.

An AI-powered interface for draw.io that allows users to generate and edit diagrams through natural language chat, supporting integration with AI agents via the Model Context Protocol (MCP).

Happy-LLM 是 Datawhale 推出的免费开源 LLM 学习教程，从 NLP 基础到 Transformer、预训练模型，再到动手搭建 LLaMA2 和训练微调，覆盖 RAG、Agent 等应用，适合有 Python 基础的学习者。

An autonomous software engineering factory that orchestrates a full team of AI agents (PMs, architects, coders, QA) to scope, build, and ship complex code changes across one or more repositories.

An Android APK that packages DeepSeek Harness to create a mobile AI agent capable of controlling system settings, apps, and web pages via Shizuku, accessibility services, and a virtual screen.

A Python SDK that provides a simplified interface for interacting with the VideoVector API to upload, process, and perform semantic searches on video content.

A free and open-source motion capture platform that provides research-grade tracking without requiring expensive, proprietary hardware.

A toolset for AI agents that provides 1,000+ pre-authenticated toolkits, session management, and authentication to enable agents to take action across various apps.

Warp is an open‑source, terminal‑based development environment that embeds AI coding agents (GPT, Claude, Gemini, etc.) and lets teams define “Warp Factories” – programmable pipelines where agents automatically triage issues, write specs, generate code, and review PRs. The client UI is built in Rust, MIT‑licensed for the UI layer and AGPL‑v3 for the rest. Install via the official binary or build locally with provided scripts, and join the community on Slack for contributions.

Tesseract is an open-source OCR engine and command-line tool that converts images of text into machine-readable text across more than 100 languages.

A high-performance distributed inference serving stack for Kubernetes that optimizes LLM production deployments across various hardware accelerators.

AI-DLC is a structured, harness-neutral framework that turns AI coding assistants into verifiable, audited software delivery workflows with human approval gates and multi-stage lifecycle management.

A set of Claude Code/Gemini‑CLI plugins that turn design knowledge into machine‑readable skills and ready‑to‑run commands, letting AI agents execute full design workflows (research, strategy, UI, interaction, systems, ops, critique) directly from the assistant.

A high-performance proxy that converts Google AI Studio's web protocols into standard OpenAI, Anthropic, and Gemini-compatible APIs, supporting multi-account management and advanced AI features.

Isaac Lab is a GPU-accelerated, open-source robotics simulation framework built on NVIDIA Isaac Sim, designed to simplify and scale robot learning tasks like reinforcement learning and motion planning.

Task‑Observer is a Claude‑specific meta‑skill that watches your AI sessions, logs corrections and repetitive patterns, and proposes updates or new skills. Users review the suggestions and install the improved skill bundles, enabling a self‑refining skill library.

A local-first multi-agent orchestration desktop app for macOS that coordinates multiple AI coding agent CLIs in one workspace with human-approval gates for workspace writes.

A coding-agent skill that automatically generates complete Chinese software copyright application materials (application form info, operation manual, and code excerpts) from your real project source code, locally and for free. It handles the official 30/30-page code rule, produces DOCX/TXT files, and lets you confirm every key detail before finalizing — no need to pay third-party document services.

Robin is a Streamlit‑based, Docker‑ready application that lets you search the dark web through Tor, automatically filters and scrapes results, and uses LLMs (OpenAI, Claude, Gemini, etc.) to generate concise investigation reports and conversational follow‑ups. It supports multiple model providers, can be run as an MCP server for integration with AI agents, and includes safety checks for scraped text. Install Tor, provide an API key, run the Docker image, and access the UI at localhost:8501.