Trending AI projects

Runonweb.ai is a free, open-source set of AI modules that run in the browser, offering one import per task (speech-to-text, background removal, translation, and more). All processing happens on-device via WebGPU or WASM, with no data leaving the device, and each module is built on an open model with credits and licenses provided.

A curated collection of production-ready skills for AI coding agents like Claude Code and Cursor, providing specialized workflows for professional web design, video presentations, and image generation.

Concord MCP is an open‑source, local‑first coordination layer that lets multiple AI coding assistants (Claude Code, Codex, Cursor, Gemini CLI, Grok Build, Goose, GitHub Copilot CLI, etc.) share a common work‑state. It provides a tiny MCP server backed by SQLite, a set of tools (`start_work`, `inspect_work`, `update_work`, `transfer_work`, `finish_work`) and a CLI (`concord status`, `dashboard`, `check`, …) so agents can claim files, detect overlaps before editing, exchange prompts/replies, hand off tasks, and generate review packets. Setup is a single `npm install -g @concord-ai/concord-mcp && concord setup`. The project is MIT‑licensed and sends optional anonymous telemetry.

An AI-powered web scraping library that uses LLMs and graph logic to extract structured data from websites and local documents using natural language prompts.

An AI-powered real-time monitoring system for the Goofish marketplace that uses multimodal models to analyze listings and notify users of matching items.

AionUi is a free, cross‑platform desktop app that turns LLM APIs into autonomous AI agents capable of reading/writing files, generating Office documents, browsing the web, and running scheduled tasks. It includes a built‑in agent engine, supports dozens of external CLI agents, offers a Team mode for coordinated multi‑agent work, and provides remote access via WebUI or chat‑platform bots.

A marketplace of AI plugins and structured workflows that encodes proven product management frameworks into AI assistants to help PMs make better product decisions.

AutoClip is an open‑source desktop/CLI/Docker app that automatically trims long videos into short, platform‑ready clips (TikTok, Douyin, YouTube Shorts, etc.) using AI transcription and LLM‑driven highlight selection, all rendered locally with optional cloud‑based model services.

A Model Context Protocol (MCP) server that enables AI agents to interact with GitHub repositories, issues, pull requests, and CI/CD workflows.

CowAgent is an open‑source, self‑hosted AI assistant that plans tasks, runs tools, keeps a personal knowledge base, and can operate as a team of cooperating agents. It supports many LLM providers, multimodal input, dozens of chat channels (Telegram, Slack, WeChat, etc.), a plug‑in skill system, and a three‑tier memory with nightly “Deep Dream” distillation. Install via a one‑line script or Docker, manage everything through a web console, and extend the system through the public Skill Hub or custom tool integrations.

AgenticX is a complete platform for building and running AI agents, offering a Python SDK, a local desktop workspace, and an enterprise-ready gateway for governance and compliance.

A multi-platform Flutter and Go application for browsing and downloading short dramas and films from various sources, performing all parsing and playback locally on the device.

A proxy tool that enables NVIDIA DLSS Frame Generation and optimized Super Resolution on RTX 20 and 30 series graphics cards by bypassing hardware architecture checks.

CopilotKit is an SDK that connects AI agents to user interfaces, enabling developers to build agent-native applications with generative UI, shared state, and multi-platform support.

A collection of over 125 specialized skills for AI agents to interact with South Korean digital services, covering everything from transport and real estate to government notices and financial data.

DLSS‑NR on AMD is a Windows‑only tool that re‑implements NVIDIA’s DLSS 5 neural‑rendering pipeline for AMD Radeon 6000‑9000 GPUs. It lets DirectX 12 games that support DLSS, FSR, or XeSS use DLSS‑style up‑scaling, ray reconstruction, and frame generation on AMD hardware via a mod‑loader and an overlay UI.

Mantis is an open‑source, AI‑driven toolkit that automates security reviews of code. It uses LLM agents to build threat models, generate exploit tests, and suggest patches, all within a sandboxed, configurable workflow. The project is experimental and intended only for isolated environments; all findings require manual verification.

LocalAI is an open‑source, modular server that lets you run LLMs, vision, speech‑to‑text, text‑to‑speech, image and video models on any hardware. Backends (llama.cpp, vLLM, whisper.cpp, etc.) are pulled as separate OCI images only when needed, providing OpenAI‑compatible APIs, multi‑user auth, privacy‑first on‑prem inference, and built‑in autonomous agents. It works on CPUs, NVIDIA/AMD/Intel GPUs, Apple Silicon, Vulkan and Jetson, and includes distributed scaling and a web UI/CLI.

A curated collection of 500+ AI agent projects with working code, framework comparisons, and industry use cases — a practical index for developers, researchers, teams, and students.

A persistent memory layer for AI coding agents that captures and compresses session data to eliminate the need for re-explaining project context across sessions and tools.

A Python‑based Agent Skill that extracts books, PDFs, slides, and other documents into a traceable knowledge base and interactive HTML learning page, supporting quick overviews or systematic study with fine‑grained source mapping and reusable method libraries.

An end-to-end document parser that converts page images into structured Markdown, including LaTeX formulas and HTML tables.

A production-grade gateway that converts AI subscriptions into stable APIs by running official client processes inside isolated, hardware-emulated virtual machines with dedicated network egress.

Null Motion is a free, local tool that shows how a finished motion-graphics ad was built by displaying the final film on top and its rough black-and-white drafts underneath, all synced frame by frame. It runs with Node 22+ (no install, no API keys), ships with three example references, and can export the whole breakdown as a single MP4 via Chrome or Edge. It is not the same as Null Studio, a separate hosted product with additional features like captions and vertical clips. The tool works by splitting references into sections at scene changes, generating simple HyperFrames drafts for each section, syncing them to the film's clock, and encoding the export with WebCodecs. It supports importing your own references via a Python script (with optional FFmpeg), and the codebase is MIT-licensed with zero runtime dependencies.

DramaClaw is a self‑hosted, source‑available AI‑driven production platform that turns scripts into short films. It offers a node‑based “infinite canvas” (XiaHua) for free‑form generation of images, video, audio and 3‑D sets, plus a structured “Series” pipeline (XiaJi) that ingests a manuscript, plans episodes, writes scripts, creates story‑boards, synthesises voice‑overs, composes video and exports the final cut. An integrated AI assistant (Xia Director) can drive both tracks. The system runs via Docker, routes all heavy model inference through an OpenAI‑compatible gateway (official or custom), and is licensed under Elastic 2.0, allowing commercial use with a required credit badge.

PixelBoard is a modified version of Gboard that enables advanced AI features like natural voice dictation, AI writing assistance, and custom UI elements on Android devices without requiring root access.

OpenHuman is an open‑source Rust‑based AI‑agent harness offering a desktop app and a library. It runs hundreds of agents on a cheap VPS by compressing prompts, using a tiny tool‑search model, and keeping the core in a single compiled process. Benchmarks show it’s the fastest, cheapest, and most memory‑efficient among seven comparable harnesses. Features include built‑in memory, voice agents, real‑browser/desktop control, 119 OAuth integrations, 14 messaging channels, and support for local or BYOK models. Install via pre‑built binaries or a simple script; developers embed it with a few lines of Rust code.

An AI agent command-line tool that automates complex professional workflows like coding and deep research using multi-agent graphs and a persistent memory system.

A Claude Code skill that automates Generative Engine Optimization (GEO) audits to help websites improve their visibility and citation readiness in AI-powered search engines.

QMD (Query Markup Documents) is a local, on‑device search engine for markdown notes, transcripts, and docs. It fuses BM25 keyword search, vector semantic search, and LLM re‑ranking (via node‑llama‑cpp). You manage collections via the `qmd` CLI, generate embeddings, and query with keyword (`search`), semantic (`vsearch`), or hybrid (`query`) modes. An optional MCP server (stdio or HTTP) lets AI agents call QMD programmatically, and a TypeScript SDK provides the same functionality inside Node/Bun apps.

A collection of human-reviewed instruction sets (skills) for AI agents to ensure they produce idiomatic, secure, and high-performance Go code.

bb is an agent‑driven IDE that can control, customise, and automate itself via desktop, web, CLI, and HTTP API surfaces. It runs tasks in live threads, supports plugins, and ships as a native desktop app (macOS stable, Linux/Windows experimental) or via `npx bb-app`. Production usage sends anonymous telemetry; development mode does not. The project is actively developed with a stable core but evolving workflows.

camofox-browser is a Node‑JS server that runs a fingerprint‑spoofed Firefox (Camoufox) and offers a REST API for AI agents to browse the web undetected. It returns compact accessibility snapshots, supports stable element refs, cookie import, proxy/GeoIP, tracing, and can be deployed via Docker, Fly.io, or Railway.

UniMate is a diffusion‑based text‑to‑animation model that can generate motion for any rigged 3‑D skeleton (humans, animals, robots, etc.) without per‑skeleton fine‑tuning. It is trained on the UniML3D dataset (13.8 k clips, 7.4 k diverse skeletons) and provides full training, data‑processing, and inference pipelines, plus a tool to bring your own GLB/FBX/GLTF rigs into the system.

Open‑source core of X’s For You timeline: candidate generation (Thunder, Phoenix, SimClusters), transformer‑based scoring, configurable safety/visibility filters, and a blending layer for ads and suggestions. Includes a transparency tool and synthetic data generation for proof‑of‑concept training.

A collection of prompts, documentation templates, and configuration recipes designed to help AI agents manage software projects with better context, quality control, and consistency.

A high-performance Rust library that converts various office document formats into clean, consistent Markdown for LLM ingestion and agentic workflows.

BKN Foundry is the backend framework for OpenBKN, an ontology‑driven platform that turns scattered enterprise data and rules into a structured knowledge graph. It provides a Context Loader to fetch only the most relevant facts (cutting token use by ~30 % and boosting accuracy), and a safety layer that validates every agent action against explicit Object‑Action‑Rule‑Constraint definitions, with full audit trails and risk simulation. Users interact via the `openbkn` CLI/SDK (TypeScript/Python) or by adding the `openbkn` skill to LLM agents. Deployed on Kubernetes, it offers unified data access, fine‑grained permissions, and traceable execution for trustworthy AI assistants in business settings.

An iterative optimization framework for Agent Skills that uses a 9-dimensional rubric and a ratchet mechanism to ensure skills only evolve toward higher quality.

A unified CLI for all Google Workspace APIs that dynamically generates its command surface, providing structured JSON output for both humans and AI agents.

An open-source, on-device AI agent for iOS and Android that integrates a sandboxed Linux shell and deep system access to perform complex mobile tasks.

Anti‑Defensive Writing is a lightweight LLM skill that rewrites academic/technical prose to remove apologetic language, redundant disclaimers, and stacked hedges while preserving evidence, methodological limits, and ethical boundaries. It consists of a single `SKILL.md` rule file loaded by agents like OpenAI Codex or Anthropic Claude Code, a zero‑dependency installer, and example before‑after rewrites. The skill follows a five‑stage pipeline (diagnosis, pruning, positive projection, uncertainty calibration, structural realignment) and can be invoked via a simple `$anti-defensive-writing` command. It’s useful for researchers, grant writers, and technical writers who want more assertive, claim‑forward writing without sacrificing scientific rigor.

A multi-agent desktop harness that coordinates various terminal-based AI coding assistants into a single, managed team with shared memory and visual oversight.

dsh-purge is a non-profit cybersecurity red-team plugin for the official DeepSeek Harness. It patches a user's own local install so host-layer prompt blocks can be separated from genuine model refusals, enabling authorized robustness testing and showing which default policies are too strict. This fork keeps the plugin working on the 0.2.1-rc.2 target.

Graphiti is an open-source framework for building temporal knowledge graphs that track how facts and relationships change over time, enabling AI agents to maintain evolving, context-aware memory.

FreeVideo is an open‑source desktop app that runs the MiniMax H3 text‑to‑video model locally via a ComfyUI plugin. It adapts precision and streams weights so it can generate videos on consumer GPUs with as little as 8 GB VRAM, offering four quality levels, multimodal inputs, LoRA support, and AI‑agent‑driven workflows.

Atticus-mac is a free, open-source voice assistant for macOS that lets you control your Mac by speaking: it can move windows, find files, read and click on screen content, dictate text, run background agents, and more. It works with Claude, Devin, Pi, or a local model, and keeps your voice private with on-device speech.

A local interface patch for Claude Desktop that adds support for Simplified and Traditional Chinese across macOS, Windows, and Windows.