Trending AI projects

Evot is a lightweight terminal‑based coding assistant that uses a LLM to read, edit, run, and write code via four built‑in tools. It works with hosted or self‑provided models, keeps prompts tiny (≈1 k tokens) using a pruning system, and adds productivity features such as background commands, looped monitoring, scheduled tasks, and shareable read‑only sessions.

A gamified browser extension that unifies user script management and web enhancement tools like AI script generation, content filtering, and media control.

MASFactory is a Python framework that lets you build, visualise, and run LLM‑based multi‑agent workflows as graphs. It supports natural‑language‑first design (Vibe Graphing), code‑first declarative/imperative authoring, and visual drag‑and‑drop editing via a VS Code extension, all feeding a unified runtime with live tracing and human‑in‑the‑loop support.

A local Model Context Protocol server that enables AI assistants to safely read and modify Obsidian vaults by interacting directly with Markdown files.

GalaxeaVLA is a vision-language-action model for general-purpose robot control that integrates reasoning and action into a single autoregressive stream for complex manipulation tasks.

Avernet is an open‑source platform that lets multiple AI agents—running on different runtimes or existing bot services—join a single, coordinated network. It provides identity, authentication, discovery, routing, and workflow tools (TaskGuard) plus diagnostics/evolution (AgentEvolve). Designed for enterprises needing persistent, governed multi‑agent systems, it ships with a Rust‑based Bot Coordination Service, web workbench, and plugins (e.g., DeepSeek Harness). Core networking features are public; advanced permission, audit, and large‑scale clustering are still in progress.

An MCP server that gives Claude private, local-first journaling and semantic search capabilities to maintain persistent memory across conversations.

Vime is an open‑source framework that combines Megatron‑LM training with vLLM inference to enable high‑performance reinforcement‑learning fine‑tuning of large language models. It offers a unified data‑flow loop, flexible custom generation hooks, and support for many modern LLMs (Qwen, DeepSeek, LLaMA 3).

A research canvas application that combines real-time search and an agentic interface to simplify complex research tasks through human-in-the-loop interaction.

A closed-loop dataset and benchmark for autonomous driving systems using the CARLA simulator to evaluate driving scores, success rates, and specific driving abilities.

FDEOps is a Node‑JS CLI that supplies a library of 35 “task‑skill” prompts for forward‑deployed engineers. Integrated with an existing AI coding assistant, it structures strategy, architecture and engineering work, stores all customer context in local Markdown files, and runs entirely offline.

awesome-hand-pose-estimation is a curated list of research papers, datasets, and resources on hand pose estimation. It organizes entries by source (arXiv, journals, conferences), year (2014–2025), and topic (datasets, workshops, challenges, theses). Each entry links to the paper's PDF and sometimes to code or project pages. It is a reference list, not a software tool, and invites community contributions.

DeepSpeed is a PyTorch‑based open‑source library that provides memory‑efficient optimizers, multi‑dimensional parallelism, and hardware‑offload engines to train extremely large deep‑learning models (up to hundreds of billions of parameters) on GPUs, CPUs, and specialized accelerators.

UltraRAG is an open‑source, low‑code framework for building, debugging, and deploying Retrieval‑Augmented Generation pipelines. It uses the Model Context Protocol to treat each component (retriever, LLM, etc.) as a plug‑and‑play server, orchestrated by YAML configuration. A built‑in web IDE provides visual pipeline editing and one‑click conversion to interactive chat demos. Install via the fast `uv` package manager or Docker, and use it for research experiments, product prototypes, or enterprise RAG services.

MiroFlow is an open-source research agent framework for multi-step internet research and complex reasoning tasks, such as future event prediction.

A curated, community-driven list of resources (articles, papers, tools, CTFs) about the prompt injection vulnerability in LLM applications, aimed at understanding, testing, and mitigating the risk.

free-code is a fork of Anthropic's Claude Code terminal AI coding assistant. It removes all telemetry (no data sent home), strips injected security-prompt guardrails (while keeping the model's own safety training), and unlocks 54 experimental feature flags. It supports five AI providers (Anthropic, OpenAI Codex, AWS Bedrock, Google Vertex AI, Anthropic Foundry), is built with Bun and TypeScript, and includes an IPFS mirror so the code persists even if the repo is taken down.

A historical archive for the OWASP Top 10 for Large Language Model Applications, providing access to legacy documentation and redirecting users to the current active security project.

A curated, structured catalog of research papers and blogs on neural-network models that compute by looping — reusing the same learned layer or operator across depth or time. It organizes 354 papers into a topic map, offers an interactive browser for searching and filtering, and tracks latest additions, citations, and open problems.

Muxin is a native cross‑platform desktop app that lets you monitor and control a personal fleet of computers via SSH/Tailscale, run AI‑agent sessions, share desktops, sync files, and manage self‑hosted GitHub Actions runners—all exposed through a JSON‑based MCP server for LLM integration.

Procedura turns text prompts or images into editable, parametric 3D models by generating OpenSCAD code through an agentic LLM pipeline.

imajev is an open‑source family of 2 B‑9 B multimodal models that take a photo (or two) plus structured text data and return typed, probabilistic decisions with an explicit *can’t‑tell* flag. Built on Qwen‑3.5 with a LoRA adapter, the models run locally (GPU or Apple‑silicon), are Apache‑2.0 licensed, and achieve top‑ranked scores on JevBench, Image JevBench, and DecisionBench. Ideal for automating high‑volume, well‑defined business decisions such as product‑photo verification, return validation, ticket routing, and safety checks, while deferring uncertain cases to a human.

Cortex turns API specifications and Markdown into typed SDKs, interactive documentation, and MCP servers for AI agents from a single project configuration.

An AI agent skill that enables one-sentence media downloads from platforms like YouTube, Bilibili, and WeChat, featuring automatic engine updates and file verification.

Confucius4-R2T2 is a low-latency, high-accuracy streaming speech recognition model that provides stable, append-only transcriptions to prevent text revisions in real-time applications.

An evaluation framework and technical report exploring the use of GPT-6 Astra as an embodied robotic policy, comparing direct control against a hybrid approach that corrects a base policy's actions.

A containerized bug bounty hunting framework that integrates 50+ security tools and an AI-agent-ready MCP server to automate reconnaissance and vulnerability assessment workflows.

Amuse is a local, .NET-based AI platform that integrates image, video, audio, and text generation pipelines into a single interface with support for multiple GPU backends.

A Python project that scrapes NBA stats and sportsbook odds, builds matchup features, trains XGBoost and TensorFlow models to predict game winners and over/under totals, and outputs expected value and Kelly‑criterion stake sizes. Includes scripts for data collection, model training, a command‑line predictor, and a simple Flask web UI.

A software development kit for Unitree robots that allows developers to build and install custom applications for robot control.

A comprehensive SEO audit tool that scans websites against 332 rules to provide actionable reports on technical SEO, performance, and AI search readiness.

llmman is a CLI/daemon that runs any LLM‑style agent (Claude Code, Codex, Copilot, etc.) on any model—local or hosted—by treating models as OCI container images. It supports multiple back‑ends (llama.cpp, vLLM, sglang, mlx‑lm), provides a unified Ollama/OpenAI/Anthropic API, enables registry‑to‑registry model transfer, signing with cosign, and aggregation of several machines into one endpoint.

Anki MCP Server is a Node‑JS bridge that implements the Model Context Protocol, exposing 53 JSON‑based tools (deck, card, note, media, model, and GUI operations) so AI assistants like Claude Desktop, Cursor, or cloud‑based ChatGPT can read, edit, and review Anki flashcards. It works via STDIO (for desktop AI clients) or HTTP, and offers a managed tunnel for remote access. Install via npm, an MCPB bundle, or from source, and configure the client to launch the server as a subprocess.

A lightweight, Kubernetes-native LLM inference platform that provides modular control and data planes for scalable, cost-efficient model serving.

A conversational AI interface that consolidates market data, technical analysis, and Interactive Brokers portfolio management into a single chat-based workflow.

OpenAccountants is an open‑source, AI‑ready tax knowledge library. It stores accountant‑reviewed or source‑cited tax guides for 231 jurisdictions, generates nightly bundles, and serves them via a Python MCP server. LLMs can query the MCP endpoint to get cited, up‑to‑date tax answers, with reviewer attribution, helping reduce hallucinations in tax‑related queries.

Webcmd is a Node‑based CLI that lets LLM agents control a real browser (via Playwright) and automatically builds a local sitemap memory of visited pages, actions, and APIs. The memory lets later agents reuse navigation knowledge, cutting token usage and cost by up to 90%. It supports profiles (cookie jars) and isolated sessions, works on research, social, shopping, and AI‑tool sites, and shows strong benchmark results on a 100‑task browser automation suite.

A low-code AI development tool based on PaddlePaddle that provides pre-trained model pipelines for rapid training, inference, and deployment across diverse hardware.

An open-source platform for securely running and managing Model Context Protocol (MCP) servers using container isolation and centralized governance.

A collection of extensible skills for AI coding agents that enable capabilities like web search, browser automation, and Google Workspace integration.

Xybrid provides native, offline SDKs (Flutter, Swift, Kotlin, React Native, Unity, Rust, Python) to run LLMs, ASR, TTS and vision models on‑device, with multi‑model pipelines and tool‑calling support.

An open-source AI slide studio for Codex coding agents that transforms prompts, repos, or files into professional presentations through a steerable, live-editing workflow.

A curated registry of AI agents that implement the Agent Client Protocol (ACP) and support user authentication, providing a discovery mechanism for client applications.

A stateless CLI tool that unifies web search, code search, and documentation scraping into a single interface designed for both human researchers and AI agents.

An agentic AI pipeline that transforms a simple idea or script into a complete cinematic micro-drama video by automating story development, character design, and video production.

A framework for building federated AI systems that enables decentralized model training across various ML frameworks and devices.

mcpc is a universal command‑line client for the Model Context Protocol (MCP). It maps every MCP operation—tools, prompts, resources, tasks, skills, OAuth authentication, and even experimental x402 payments—into intuitive shell commands. Designed for AI agents and developers, it lets you inspect, debug, and script MCP workflows using standard Unix tools (e.g., jq, xargs). Features include persistent named sessions, progressive tool discovery via `grep`, JSON‑only output mode, secure OAuth 2.1 storage (OS keychain or protected file), AI sandboxing, and lightweight cross‑platform support.

Hyper-Extract is a CLI tool that transforms unstructured documents into structured knowledge graphs and databases, enabling precise querying, visualization, and incremental updates.