Trending repos

TensorFlow is a widely-used open-source ML framework with 195k+ stars. It remains foundational infrastructure for deep learning across academia and industry.

Agent harness optimization system for Claude, Codex, and other LLMs with skills, memory, security, and research-focused development. Appears to be a developer tool for agent orchestration and performance tuning across multiple AI platforms.

AutoGPT is an open-source framework for building autonomous AI agents with a focus on accessibility and extensibility. With 184K stars, it enables developers to create and customize agentic systems without deep AI expertise.

Ollama is a popular open-source framework for running LLMs locally, supporting models from DeepSeek, Qwen, Gemma, and others. With 171k+ stars, it remains a central tool in the OSS LLM ecosystem for easy model deployment.

Community-driven prompt repository with 162k+ stars for sharing and discovering ChatGPT, Claude, and other LLM prompts. Offers self-hosted deployment option for organizational privacy.

Hugging Face Transformers is the de facto standard library for deploying and fine-tuning state-of-the-art models across text, vision, audio, and multimodal domains. With 160k+ stars, it's the essential infrastructure that most researchers and practitioners use for both inference and training.

Hermes Agent is an open-source AI agent framework that integrates with major LLM providers (Claude, ChatGPT, Anthropic). It's trending on GitHub with 160k stars, suggesting substantial developer adoption for building scalable autonomous agents.

Langflow is a Python framework for building and deploying AI agents and workflows with visual flow design. A well-established open-source project with 148K+ stars, it enables developers to compose LLM-based applications without extensive coding.

Dify is a production-ready platform for building and deploying agentic AI workflows, with 142K+ stars on GitHub. It supports multiple LLM providers and streamlines agent development from prototyping to production.

Open WebUI is a user-friendly web interface for running and interacting with LLMs locally (via Ollama) or through APIs (OpenAI, etc.). With 138K stars, it's a mature, widely-adopted tool for developers and non-technical users to work with language models.

LangChain is a leading open-source framework for building agent-based AI applications, with 137K+ stars. It provides abstractions for orchestrating LLM chains, tools, and memory—widely used for enterprise AI systems.

Firecrawl is a web scraping and data extraction tool designed for AI agents, converting web content to clean, AI-ready formats. With 122K+ stars, it's a popular infrastructure project for building web-connected AI systems.

A curated collection of 100+ open-source AI agent and RAG applications ready to clone and deploy. Comprehensive resource for developers building with LLMs and retrieval-augmented generation.

Microsoft's comprehensive 21-lesson curriculum for building generative AI applications, covering foundational concepts through practical implementation with GPT, DALL-E, and Azure tools. A well-maintained educational resource that has gained significant community traction for learning hands-on GenAI development.

PyTorch's main repository now at 100k+ stars. Core deep learning framework with GPU acceleration, dynamic computation graphs, and automatic differentiation.

Educational repository implementing a ChatGPT-like LLM in PyTorch from scratch with step-by-step guidance. Widely popular (95k+ stars) for learning transformer architecture, attention mechanisms, and fine-tuning fundamentals.

Browser-use is a Python framework enabling AI agents to interact with websites through browser automation. It bridges the gap between LLMs and web tasks, allowing agents to autonomously navigate, extract data, and perform actions online.

Real-time face-swap and deepfake generation tool using a single reference image, with webcam integration. Trending open-source project with 93K stars demonstrating accessible deepfake creation.

Curated list of MCP (Model Context Protocol) servers with 87k+ stars. Useful reference for developers building Claude integrations and AI tools that use the standard.

OpenCV is a mature, widely-used open-source library for computer vision and image processing with 87k+ stars. It remains a foundational tool for researchers and developers building vision applications.

Official Model Context Protocol servers repository containing reference implementations and tools for building MCP-compatible applications. This is the core infrastructure enabling AI agents to interact with external systems and data sources.

Microsoft's structured ML curriculum with 26 lessons, 52 quizzes, and 12-week program covering classic machine learning fundamentals. Educational resource for beginners learning ML algorithms and data science.

RAGFlow is an open-source RAG engine combining retrieval-augmented generation with agentic capabilities to provide a context layer for LLMs. With 80K+ stars, it's a mature, production-ready framework for building intelligent retrieval systems.

vLLM is a high-throughput, memory-efficient inference engine for LLMs with 80k+ stars. It's a foundational tool widely adopted for serving large models at scale in production.

Netdata is a full-stack observability platform with AI capabilities for monitoring infrastructure and applications. The tool targets lean teams with performance-optimized monitoring and alerting, now trending on GitHub with 78k+ stars.

TradingAgents is a multi-agent LLM framework for financial trading that coordinates multiple AI agents for trading decisions. Open-source Python project with strong community interest (78k+ stars) demonstrating practical LLM application in high-stakes finance domain.

PaddleOCR is a lightweight, multilingual OCR toolkit that converts PDFs and images into structured data for AI systems, supporting 100+ languages with strong integration for document parsing and RAG pipelines.

LobeHub is a platform for managing and orchestrating multiple AI agents with hiring, scheduling, and reporting capabilities. The TypeScript-based project has gained significant GitHub traction (77k+ stars) for unified AI agent operations.

claude-mem enables persistent memory across AI agent sessions by capturing, compressing, and reinitializing context for future interactions. Supports Claude, OpenAI, Gemini, and other models with integration for popular agent frameworks.

OpenHands is an AI-driven development framework enabling autonomous agents to write, debug, and deploy code. With 74k+ stars, it's a mature open-source platform for building AI-powered developer tools across multiple LLM backends.

LlamaFactory is a unified framework for efficiently fine-tuning 100+ LLMs and VLMs, supporting popular models like GPT, Gemma, and DeepSeek. It enables instruction-tuning and agent development with a single codebase.

ByteDance releases Deer-Flow, an open-source agentic framework enabling multi-step reasoning over extended horizons with sandboxes, memory, tools, and sub-agents for research, coding, and creative tasks.

OpenBB is an open-source financial data platform designed for analysts, quants, and AI agents, offering access to equity, derivatives, crypto, and fixed-income data. With 67K stars, it's a widely-adopted toolkit for building finance-focused AI applications and quantitative workflows.

gpt4free is a Python library providing free access to multiple frontier language models (GPT-5.3, Claude Opus 4.6, DeepSeek v3.2, Gemini 3, Kimi 2.5) through unified APIs. Popular open-source project (66k stars) that abstracts away vendor-specific integrations.

scikit-learn is a foundational Python ML library offering classical algorithms for classification, regression, clustering, and dimensionality reduction. It remains essential infrastructure for data science workflows despite being a mature, well-established project.

Crawl4AI is an open-source web crawler and scraper optimized for LLM workflows, enabling efficient extraction of web content for AI model training and reasoning. With 66k+ stars, it's becoming a foundational tool for building LLM-powered applications that need reliable web data.

Microsoft's educational repository with 12 lessons for building AI agents, covering foundational agentic patterns, RAG integration, and frameworks like AutoGen. A practical learning resource for developers entering the agent-building space.

Unsloth Studio is a web-based UI for locally training and running open-source models including Gemma 4, Qwen3.6, and DeepSeek. Enables efficient fine-tuning and inference without cloud dependencies.

MinerU converts PDFs and Office documents into LLM-ready markdown/JSON, enabling seamless integration with agentic workflows. High-star repo addressing document extraction at scale for AI applications.

Keras is a widely-used deep learning framework that abstracts over TensorFlow, PyTorch, and JAX backends. It remains a foundational tool for researchers and practitioners building neural networks with a high-level, intuitive API.

Pathway is a Python ETL framework for stream processing, real-time analytics, LLM pipelines, and RAG applications. With 63K+ stars, it's a mature open-source project for building data pipelines that integrate with AI workflows.

A Claude Code skill that reduces token usage by 65% through caveman-style simplified language. Demonstrates creative prompt engineering for cost optimization with a humorous approach.

A minimal Claude-based agent framework built from scratch, demonstrating core patterns for agentic AI systems. Educational resource with 61K stars showing practical implementation of agent harnesses using Claude.

AnythingLLM is an open-source AI productivity platform supporting local and remote models with privacy-first design and minimal setup. With 60k+ stars, it's a popular choice for building custom AI agents without complex configuration.

TrendRadar is an AI-driven public opinion and trend monitoring tool that aggregates multi-platform content, RSS feeds, and applies intelligent filtering, translation, and sentiment analysis. It supports MCP integration, local/cloud data storage, and notifications via WeChat, Feishu, Dingtalk, Telegram, email, and other channels.

Meilisearch is a Rust-based search engine API offering hybrid search with AI capabilities for web and app integration. Widely adopted (57K+ stars) with strong enterprise feature set including fuzzy matching and faceting.

Python tool that generates short videos automatically using AI language models with a single click. Combines LLM capabilities with video processing for rapid TikTok-style content creation.

Ultralytics YOLO is a widely-adopted open-source framework for real-time object detection, instance segmentation, and image classification. With 57k+ stars, it's a foundational tool used across industry for computer vision tasks.

Mem0 is a universal memory layer framework for AI agents that enables persistent context and learning across conversations. With 56k stars, it's become a popular infrastructure tool for building stateful AI applications.

Context7 is a platform providing real-time code documentation tailored for LLMs and AI code editors, enabling better context awareness during code generation and editing tasks.

Deepfakes/faceswap is a widely-used open-source Python framework for face-swapping and deepfake generation. With 55K+ stars, it remains a prominent tool in synthetic media, relevant for understanding both creative and adversarial deepfake capabilities.

Ruflo is an agent orchestration platform for Claude enabling multi-agent swarms, autonomous workflows, and conversational AI with RAG integration and native Claude tooling. Early traction with 53K stars suggests meaningful adoption in the agentic systems space.

Flowise is a visual builder for constructing AI agents and workflows without code, supporting agentic AI patterns and integrations with LangChain. With 52k stars, it's a mature open-source platform enabling rapid prototyping of chatbots and agent systems.

MemPalace is an open-source AI memory system that claims best-in-class benchmarking performance. It integrates with ChromaDB and MCP, offering persistent context management for LLM applications.

Tutorial repository teaching agent fundamentals from scratch, covering principles and practical implementation for LLM-based agents with RAG integration. Growing educational resource for developers learning agent development.

RTK is a Rust CLI proxy that cuts LLM token usage by 60–90% on development commands through intelligent caching and compression. It's a zero-dependency binary gaining rapid adoption for cost-effective AI-assisted coding workflows.

Open-source framework for training a 64M-parameter LLM from scratch in ~2 hours, with 50k+ GitHub stars. Demonstrates efficient training pipeline and serves as educational resource for understanding LLM training fundamentals.

Graphify converts code repositories, schemas, and documentation into queryable knowledge graphs compatible with Claude, Codex, Cursor, and other AI assistants. Enables semantic search and context retrieval across mixed code/data assets.

Local-first, open-source design tool that generates web, desktop, mobile prototypes and multimedia assets using Claude and other LLMs. Supports 19 AI skills, 71 design systems, and exports to HTML/PDF/PPTX/MP4 formats.

LlamaIndex is a popular Python framework for building document agents and RAG applications with OCR capabilities. It enables developers to connect LLMs with external data sources and build multi-agent systems.

Julia is a high-performance programming language designed for numerical computing and scientific machine learning. Trending on GitHub with 48k stars, it remains a solid choice for HPC and ML workloads, though not a new release or breakthrough.

LiteLLM is a widely-used Python SDK and AI gateway that provides unified API access to 100+ LLM providers (OpenAI, Anthropic, Azure, Bedrock, etc.) with built-in cost tracking, rate limiting, and logging. With 47K+ stars, it's a key infrastructure tool for multi-model LLM applications.

oobabooga/textgen is a mature open-source desktop application for running LLMs locally with text, vision, and tool-calling capabilities, offering OpenAI/Anthropic-compatible APIs while ensuring 100% privacy. With 47K stars, it's a heavily-used community tool for local LLM inference.

Career-ops is an AI-powered job search automation system leveraging Claude Code with 14 skill modes, a Go dashboard, PDF generation, and batch processing capabilities. Addresses job search efficiency through intelligent agent-driven workflows.

LocalAI is a GPU-optional open-source engine supporting multiple model types (LLMs, vision, voice, image, video) across diverse hardware. Reached 46K+ stars as a practical option for running AI workloads locally and decentralized.

JeecgBoot is an AI-powered low-code platform combining code generation and zero-code configuration modes, enabling rapid system development through AI-driven workflows, form design, and process orchestration. Supports mainstream LLMs and claims to eliminate 80% of repetitive Java development work.

Cherry Studio is an AI productivity platform offering unified access to 300+ assistants and frontier LLMs with autonomous agents and smart chat capabilities. Built in TypeScript, it integrates with major models (Claude, DeepSeek, etc.) and has gained significant GitHub traction.

Apache Airflow is a mature workflow orchestration platform for authoring, scheduling, and monitoring data pipelines via DAGs. Widely used in data engineering but not a frontier AI advancement.

CowAgent is a lightweight Python AI agent framework supporting multiple LLMs (DeepSeek, OpenAI, Claude, Gemini, Qwen) with integrations for WeChat, Feishu, DingTalk, and other platforms. It enables autonomous task planning, OS/resource access, skill creation, and long-term memory for building personal AI assistants and enterprise digital workers.

Streamlit is a Python framework for rapidly building and deploying data apps without frontend expertise. It's a mature, widely-adopted tool in the ML/data science ecosystem with 44k+ stars.

Milvus is a mature, high-performance vector database with 44k+ stars, designed for scalable ANN search across cloud deployments. A widely-adopted foundational tool for RAG and embedding-based AI applications.

Kong is an open-source API gateway with AI capabilities, enabling management and orchestration of APIs and AI services at scale. A mature, widely-adopted infrastructure tool for cloud-native deployments.

Lightweight open-source AI agent framework supporting multiple models (Claude, ChatGPT) and tool integration for chat and workflow automation. Rapidly gaining traction with 42k+ stars, positioning as an accessible alternative to heavier agent frameworks.

Gradio is a Python framework for rapidly building and deploying machine learning web interfaces. With 42k+ stars, it's a widely-adopted tool for sharing ML models and data applications.

Jan is an open-source desktop app that runs LLMs 100% offline, offering a ChatGPT-like interface without cloud dependency. It's built with Tauri and supports various models via llama.cpp, gaining significant GitHub traction.

Ray is a distributed computing engine for ML with libraries for training, tuning, and serving. It's a mature, widely-adopted framework for scaling AI workloads across clusters.

DeepSpeed is Microsoft's open-source optimization library for efficient distributed training and inference of large language models, supporting techniques like ZeRO, tensor parallelism, and mixture-of-experts. It's a foundational tool widely used in the LLM community for scaling training to billions of parameters.

ColossalAI is a distributed training framework for large foundation models, enabling efficient scaling via data/model parallelism and heterogeneous computing. A mature, well-starred project that reduces training costs and accelerates AI model development.

Paperless-ngx is an open-source document management system that scans, indexes, and archives documents using Django and Angular. While tagged with machine-learning, it's primarily a document workflow tool rather than an AI research project.

Repository collecting leaked system prompts from major AI models including Claude, GPT-5.5, Gemini, Grok, and others. Provides visibility into instruction sets and behavioral constraints these models operate under.

PhotoPrism is a self-hosted, AI-powered photo management application for organizing and searching personal photo libraries with machine learning capabilities. It offers decentralized control over photo data with features comparable to Google Photos but running on private infrastructure.

Supervision is a Python framework for building reusable computer vision tools with support for classification, segmentation, and detection tasks. It's a widely-adopted OSS project (39k+ stars) enabling developers to rapidly prototype and deploy CV pipelines.

MindsDB's open-source platform for building production AI systems with control over deployment and customization. Focused on agents, analytics, and database integration with 39K+ stars.

LLM-driven stock analysis system for A/H/US markets combining real-time market data, news, and LLM-based decision dashboards with multi-channel alerts. Demonstrates practical application of LLMs to quantitative finance and trading.

Tesseract.js is a JavaScript OCR library supporting 100+ languages, running via WebAssembly for in-browser text recognition. Popular open-source project (38k stars) enabling optical character recognition without server dependencies.

Google Research's public repository containing implementations of published research papers and experimental code. A broad collection spanning ML fundamentals to cutting-edge work, widely used as reference implementations.

PyTorch Image Models (timm) is the largest open collection of pretrained image encoders and backbones including ResNet, ViT, EfficientNet, ConvNeXt, and 100+ architectures with training, evaluation, and export utilities. Essential infrastructure for computer vision practitioners and researchers.

Google's langextract is a Python library for extracting structured data from unstructured text using LLMs, with source grounding and interactive visualization. Enables developers to build reliable information extraction pipelines with transparency into where extracted data originated.

Multi-harness agentic plugin marketplace supporting Claude Code, Cursor, Codex CLI, and other platforms. Enables developers to build and share reusable agent skills across multiple AI coding environments.

LightRAG is an efficient retrieval-augmented generation framework accepted at EMNLP 2025 that optimizes speed and simplicity for RAG systems using knowledge graphs. It's gaining traction as a faster alternative to existing GraphRAG implementations.

MediaPipe is Google's cross-platform ML framework for live and streaming media processing, supporting audio, video, and computer vision tasks. It's a mature, widely-used production framework with 35k+ stars that enables developers to build efficient edge ML applications.

Educational hub with in-depth tutorials on LLMs, RAG systems, and AI agent applications. Growing community resource (35k stars) covering practical implementation patterns and real-world deployment strategies.

CodeWhale is a CLI-based coding agent for open-source LLMs, written in Rust with terminal UI support. Enables developers to use locally-runnable models (like DeepSeek) for code generation and assistance without cloud dependencies.

Hugging Face's Diffusers library provides production-ready implementations of diffusion models for image, video, and audio generation in PyTorch. A widely-adopted open-source framework with 33K+ stars that enables researchers and developers to work with SOTA generative models.

GitHub's curated collection of community-contributed prompts, agents, and configurations for GitHub Copilot. A growing resource (33k+ stars) showcasing practical extensions and customizations for the AI assistant.

spaCy is a mature, production-grade NLP library for Python with 33k+ stars, offering fast entity recognition, dependency parsing, and text processing pipelines. It remains a standard tool for industrial NLP applications and data science workflows.

ChatDev enables software development entirely through LLM-powered multi-agent collaboration, where AI agents take on roles like designer, coder, and tester to build full applications. The 2.0 release expands this framework for more complex project coordination.

Microsoft's GraphRAG is a modular, open-source RAG system that uses graph-based retrieval to augment LLM reasoning. With 33k+ stars, it represents a mature approach to structured knowledge retrieval for enterprise AI applications.

Microsoft released Playwright MCP, a server that integrates Playwright browser automation with the Model Context Protocol. Enables AI agents to control web browsers programmatically for testing, scraping, and interaction tasks.

Netron is a popular open-source visualizer for neural network models across frameworks (CoreML, Keras, TensorFlow, PyTorch, etc.). Widely used by ML practitioners for model inspection and debugging.

AstrBot is a Python-based AI agent framework that integrates multiple IM platforms, LLMs, and plugins—positioning itself as an open-source alternative to proprietary solutions. Useful for developers building multi-platform conversational AI without vendor lock-in.

LangGraph is a Python framework for building resilient AI agents with support for multiple LLM providers and enterprise deployment. It has achieved 32K+ stars and is widely used for agentic AI workflows.

PageIndex is a document indexing framework designed for reasoning-based RAG systems that operate without traditional vector embeddings. It enables agents to efficiently retrieve and reason over documents through alternative indexing strategies.

CopilotKit is a frontend framework for building AI agents and generative UI with React and Angular support. It provides developer infrastructure for integrating agentic AI patterns into web applications.

Qdrant is a high-performance vector database and search engine written in Rust, designed for AI applications with support for hybrid search, embeddings, and massive scale. It has gained significant adoption with 31k+ stars and offers both open-source and cloud deployment options.

PyTorch Lightning is a framework enabling seamless pretraining and finetuning of AI models at any scale (1 to 10,000+ GPUs) without code modifications. It abstracts distributed training complexity, making it valuable for researchers and practitioners working across diverse model sizes.

GitHub's official Model Context Protocol (MCP) server implementation in Go, enabling AI tools and agents to interact with GitHub APIs through the MCP standard interface.

AIHawk is an AI agent that automates job applications by tailoring submissions across multiple platforms. The project has gained significant traction (29.8K stars) as a practical tool for streamlining job hunting.

Onyx is an open-source AI chat platform supporting any LLM with enterprise search and RAG capabilities. It's gained significant traction (29.5k stars) as a modular alternative to proprietary AI assistants.

An open-source prompt optimization tool that helps users write better prompts and improve AI outputs. Now at 29K stars, it's become a popular utility in the prompt engineering community.

Curated collection of data science learning resources and real-world problem applications with 29k+ stars. Comprehensive reference covering analytics, mining, visualization, and deep learning fundamentals.

Sim is a TypeScript framework for building, deploying, and orchestrating AI agents as a central intelligence layer. It integrates with Anthropic and provides agentic workflow automation tooling.

SGLang is a high-performance serving framework for LLMs and multimodal models with 28K+ stars. It's a mature, widely-adopted inference framework supporting multiple model families and hardware accelerators.

FastGPT is an open-source knowledge platform for building RAG and question-answering systems with visual workflow orchestration, supporting multiple LLM providers. It abstracts away deployment complexity for developers building knowledge-based AI applications.

fastai is a popular PyTorch-based deep learning library (28K stars) that simplifies training neural networks through high-level APIs and interactive notebooks. It's widely used for computer vision, NLP, and tabular data tasks.

Semantic Kernel is Microsoft's SDK for integrating LLMs into applications with plugins, memory, and orchestration. It's a mature, widely-adopted framework (27k+ stars) that simplifies prompt engineering and agent building across multiple languages.

Langfuse is an open-source LLM engineering platform providing observability, metrics, evals, prompt management, and playground features with broad SDK integration (OpenAI, LiteLLM, Langchain, etc.). It's a mature YC W23 project with 27K+ stars addressing the operational side of LLM development.

Comprehensive GitHub repository demonstrating advanced RAG techniques with detailed Jupyter notebook tutorials. Covers embeddings, LangChain, LlamaIndex, and LLM integration patterns.

Label Studio is a widely-used open-source data labeling platform supporting multiple annotation types (bounding boxes, segmentation, NER, etc.) with standardized output. Essential infrastructure for ML teams preparing training datasets.

Autonomous research agent that conducts in-depth investigations on any topic using pluggable LLM providers. Supports multiple LLM backends and MCP integration, with 27k+ GitHub stars indicating substantial adoption for automated research workflows.

OpenClaude is a TypeScript CLI agent framework that runs anywhere and integrates with multiple backends. Early-stage project with 27k stars suggesting community interest in accessible Claude tooling.

InvokeAI is a mature, widely-adopted creative engine for Stable Diffusion with 27K+ stars, offering professional-grade WebUI for image generation, inpainting, and img2img workflows. Powers multiple commercial products and remains a leading open-source tool for generative art.

Comprehensive GitHub repository with thousands of resources on ethical hacking, bug bounties, DFIR, AI security, vulnerability research, and exploit development. Curated by Omar Santos with 26k+ stars; useful reference for security practitioners.

Graphiti is a Python framework for building real-time knowledge graphs that integrate with AI agents and RAG systems. It enables dynamic graph construction for LLM-powered applications seeking structured memory and reasoning.

MLflow is an open-source platform for managing the full lifecycle of LLM and ML applications, including debugging, evaluation, monitoring, and optimization. With 26k stars, it's a widely-adopted tool for teams building production AI systems at scale.

AionUi is a free, open-source local desktop app supporting 20+ AI agents and CLI tools (Claude Code, OpenClaw, Hermes, Gemini) with customizable assistant configurations. Positions itself as a unified interface for multi-agent workflows.

ScrapeGraphAI is a Python library that uses LLMs to intelligently scrape and extract data from web pages, positioning itself as an alternative to traditional crawlers. The project has gained significant traction (25k+ stars) as developers seek AI-native approaches to data extraction.

Chat2DB is an AI-powered SQL client with GUI support for 10+ databases (MySQL, PostgreSQL, Oracle, SQL Server, etc.). It lets users query databases through natural language, reducing need for SQL expertise.

Deprecated educational repository on machine learning and deep learning. The original Hands-on ML project has been superseded by newer versions (handson-ml3 and handson-mlp).

Strix is an open-source tool that uses AI agents to automatically discover and remediate vulnerabilities in applications. It combines LLM-based penetration testing with autonomous agents to help developers identify security issues.

SHAP is a widely-used library that explains ML model predictions using game theory (Shapley values). With 25k+ stars, it's a foundational tool for model interpretability across deep learning, gradient boosting, and classical ML.

Haystack is an open-source orchestration framework for building production LLM applications with modular pipelines, RAG, agents, and multimodal support. Mature project (25K+ stars) with explicit control over retrieval, routing, and generation components.

FastMCP is a Python framework for rapidly building MCP (Model Context Protocol) servers and clients with a Pythonic API. Trending on GitHub with 25K+ stars, it streamlines integration of language models with external tools and data sources.

Repomix converts entire codebases into single AI-friendly files for feeding to LLMs. Tool gained 25k stars, enabling developers to efficiently share code context with Claude, ChatGPT, and other AI models.

PyTorch implementation of Vision Transformer (ViT) for image classification using transformer encoders. A highly-starred reference repo widely used for vision transformer research and production applications.

Vercel's AI SDK is a TypeScript library for building AI applications and agents, with 24k+ stars. It provides unified abstractions for LLM integration (Claude, Gemini, etc.) and generative UI, lowering friction for TypeScript developers.

OpenViking is an open-source context database for AI agents that unifies memory, resources, and skills management through a filesystem paradigm, enabling hierarchical context delivery and self-evolution.

Harvard Edge Lab's CS249r course textbook on machine learning systems, covering cloud ML, edge computing, and embedded systems. Educational resource gaining significant GitHub attention for practical ML systems knowledge.

DeepTutor is an agent-native personalized learning assistant leveraging large language models and multi-agent systems for interactive education. It combines RAG, deep research capabilities, and CLI tooling to enable adaptive tutoring at scale.

Agent2Agent (A2A) is an open protocol for multi-agent communication and interoperability, enabling opaque agentic systems to coordinate. Addresses fragmentation in agent ecosystems by providing a standardized way for different AI agents to discover and work together.

PaddlePaddle is Baidu's open-source deep learning framework for high-performance training and cross-platform deployment, with 24k stars. Industrial-grade ML framework used at scale in production systems.

Curated ranking of machine learning Python libraries updated weekly, covering automl, deep learning, data science, and related domains. Popular community resource for discovering and comparing ML tools.

ncnn is Tencent's high-performance neural network inference framework optimized for mobile platforms, supporting ARM NEON and multiple model formats. A mature, widely-used C++ library for on-device deep learning inference with 23k+ stars.

Official Python SDK for the Model Context Protocol enabling developers to build MCP servers and clients. With 23k stars, this is the primary tooling for integrating MCP with Python applications.

Nautilus Trader is a production-grade Rust-native trading engine with deterministic event-driven architecture, supporting algorithmic trading across crypto, equities, forex, futures, and options. Gaining significant traction (22.8K stars) as a high-performance framework for systematic traders and ML-based strategies.

NocoBase is an open-source no-code platform combining AI agents with a WYSIWYG interface for rapid business system development. It leverages proven infrastructure rather than generating from scratch, targeting speed and reliability for CRM, dashboards, and data management.

Activepieces is an open-source platform for building AI agents and workflows with support for ~400 MCP (Model Context Protocol) servers. It provides a framework for connecting AI agents to tools and automating multi-step AI workflows.

OpenCLI lets AI agents interact with websites through a CLI interface using your authenticated browser session. Enables programmatic web automation without needing separate credentials.

Curated GitHub repository listing free LLM API resources including Claude, Gemini, Llama, and OpenAI endpoints. Useful reference for developers seeking cost-free inference options.

Large-scale recommendation systems repository with best practices, utilities, and examples. Widely-used open-source resource for building production recommendation pipelines with deep learning and classical approaches.

Skyvern is an open-source Python framework for automating browser-based workflows using AI and LLMs. It enables autonomous task execution across web applications without manual scripting.

Hugging Face Datasets is a widely-adopted library providing access to thousands of ready-to-use datasets with efficient data loading and manipulation tools. It's a foundational resource for ML practitioners and researchers building and training models.

Promptfoo is a testing and evaluation framework for prompts, agents, and RAG systems with red-teaming capabilities and multi-model comparison. Used by OpenAI and Anthropic, it offers declarative configs and CI/CD integration for LLM evaluation at scale.

Open-source PDF parser designed for AI workflows, converting PDFs into AI-ready formats with accessibility features and bounding-box extraction. Trending on GitHub with 21K stars, addresses a common bottleneck in RAG and document processing pipelines.

Ciphey is an AI-powered tool that automatically decrypts ciphers, decodes encodings, and cracks hashes without prior knowledge of the encryption method. Useful for CTF competitions and cryptanalysis workflows.

PEFT is Hugging Face's parameter-efficient fine-tuning library supporting LoRA, adapters, and other techniques for efficient model adaptation. With 21K+ stars, it's the dominant open-source tool for training large models on limited hardware.

marimo is a reactive Python notebook that combines reproducible experiments, SQL queries, script execution, and app deployment in a Git-friendly format with an AI-native editor. It's gained significant traction (21K+ stars) as a modern alternative to Jupyter for data science and ML workflows.

MaxKB is an open-source platform for building enterprise-grade AI agents with knowledge base integration, supporting multiple LLMs and frameworks like LangChain. The project has gained significant traction (21K+ stars) as a practical tool for agentic AI workflows.

ONNX is an open standard for representing machine learning models, enabling interoperability across frameworks like PyTorch, TensorFlow, and scikit-learn. It's a foundational infrastructure project with 20k+ stars that reduces vendor lock-in and simplifies model deployment.

Dyad is an open-source local AI app builder for power users, positioning itself as an alternative to Lovable, Replit, and Bolt. It's gaining traction with 20k+ stars and supports multiple AI backends including Anthropic and DeepSeek.

Agent-Reach is a Python CLI tool enabling AI agents to access and search multiple internet platforms (Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu) without API fees. Provides agents with real-time web visibility across diverse sources through a unified interface.

VoxCPM2 is a tokenizer-free TTS system enabling multilingual speech generation, creative voice design, and voice cloning. Released by OpenBMB, it represents a significant advancement in speech synthesis technology with strong community adoption (19.5k stars).

Opik is an open-source platform for debugging, evaluating, and monitoring LLM applications with end-to-end tracing, automated evals, and production dashboards. Integrates with LangChain and LlamaIndex, addressing a key pain point in LLM app observability.

ppt-master generates natively editable PowerPoint presentations from documents using AI, creating real vector shapes and animations instead of rasterized images. Enables programmatic PPTX generation with full Office compatibility.

GitHub repository providing structured workflows and prompts for academic research using Claude, covering literature review, writing, peer review, and revision cycles. Useful for researchers leveraging AI to streamline the full publication pipeline.

DB-GPT is an open-source agentic AI assistant for database and data product integration, supporting multiple LLMs including GPT and DeepSeek. It combines agent architecture with database connectivity for enterprise AI+data workflows.

Hugging Face's TRL library enables training transformer language models using reinforcement learning techniques. It's a widely-adopted open-source framework (18K+ stars) for RLHF and related alignment approaches.

elizaOS/eliza is an open-source agentic operating system framework in TypeScript for building autonomous agents with chat, crypto, and Discord integrations. Growing rapidly with 18K stars, it enables developers to create multi-platform AI agents.

Parlant is a Python framework for building reliable customer-facing AI agents with controlled, consistent LLM interactions. It provides an interaction control harness designed to ensure predictable agent behavior in production customer service scenarios.

Open-source comprehensive guide covering ML engineering practices including debugging, GPU optimization, inference, and LLM deployment. A community reference with 17K+ stars for practitioners building production ML systems.

WeClone lets you create a personalized AI twin by fine-tuning language models on your chat history, then deploying it as a chatbot. Captures your writing style and personality for a digital avatar.

DocsGPT is an open-source RAG platform enabling private AI agents and assistants with built-in agent builder, document analysis, and multi-model support. Useful for enterprises needing on-premise search and agent capabilities without external API dependencies.

PyVideoTrans is a Python tool that translates videos across languages with automatic dubbing and subtitle generation. It combines speech-to-text, translation, and text-to-speech to create localized video content.

Cognee is a Python library that abstracts memory management for AI agents, allowing developers to implement sophisticated context and recall systems in minimal code. It addresses a core challenge in agent development—managing and reasoning over accumulated context efficiently.

Leon is an open-source personal AI assistant written in TypeScript that automates tasks and provides chat-based interactions. With 17k+ stars, it represents a mature OSS agent project in the personal assistant space.

NVIDIA NeMo is a scalable generative AI framework for building LLMs, multimodal models, and speech AI systems (ASR, TTS). It's a comprehensive toolkit for researchers and developers across multiple modalities with 17k+ GitHub stars.

KTransformers is a framework for optimizing LLM inference and fine-tuning across heterogeneous hardware setups. It enables researchers and builders to experiment with different inference strategies without rewriting code.

Fully autonomous AI agent system in Go for conducting complex penetration testing tasks. Combines multi-agent architecture with LLM capabilities for security testing automation.

CAMEL is an open-source multi-agent framework for building AI systems where agents communicate and cooperate. It's trending with 17K+ stars and focuses on understanding how agent systems scale.

Open-source infrastructure for computer-use agents with sandboxed environments, SDKs, and benchmarks for training AI agents that control full desktops across macOS, Linux, and Windows.

Google Cloud's official repository of sample code and notebooks for building generative AI applications on GCP, including integration with Gemini Enterprise Agent Platform. Provides practical examples for developers adopting Google's AI services.

Open-source self-hosted platform aggregating 200+ generative AI models (Flux, Midjourney, Kling, Sora, Veo) for image and video creation without content filters. MIT-licensed alternative to commercial AI studios.

Megatron-LM is NVIDIA's framework for training large transformer models with distributed parallelism techniques. Core infrastructure used by many labs for scaling LLM training.

Open-source book repository covering mathematical foundations of reinforcement learning with tutorials and course materials. Gaining traction as a comprehensive educational resource for RL theory.

Weaviate is a mature open-source vector database combining vector search with structured filtering, HNSW indexing, and hybrid search capabilities. Well-established project (16k+ stars) serving production ML/search use cases.

Microsoft's open-source curriculum teaches Model Context Protocol (MCP) fundamentals through practical cross-language examples (.NET, Java, TypeScript, JavaScript, Rust, Python). It covers building modular, scalable AI workflows from setup to orchestration.

Production-grade platform for building agentic IM bots across Discord, Slack, WeChat, Telegram, Dingtalk, and 10+ other messaging platforms. Includes agent orchestration, knowledge base management, and plugin system with integrations to ChatGPT, Claude, DeepSeek, Dify, and other LLM/workflow providers.

Transformers.js enables running Hugging Face transformer models directly in web browsers without server dependency. Mature project (16K+ stars) providing accessible ML inference for web developers.

Collection of 313+ Claude skills and plugins for coding agents across Claude Code, Cursor, Gemini CLI, and others, covering engineering, marketing, compliance, and business operations use cases. Popular GitHub project with 15K+ stars enabling developers to extend AI coding assistants with domain-specific capabilities.

DeepEval is an open-source framework for evaluating LLM outputs using customizable metrics and automated testing. It helps developers measure quality, hallucination, bias, and other dimensions of LLM performance with minimal setup.

Memvid is a memory layer for AI agents that replaces RAG pipelines with a serverless, single-file system for instant retrieval and long-term memory. It's gaining traction as a practical alternative to complex retrieval architectures.

WrenAI is an open-source framework that equips AI agents with grounded memory, context, and SQL access across 20+ databases to build agentic analytics, text-to-SQL, and GenBI applications. It abstracts away data plumbing so agents can reason over real data sources reliably.

Curated collection of GPT-Image-2 API documentation, integration examples, and prompt templates for image generation. Helps developers quickly implement and optimize image generation workflows.

Google released an open-source MCP server enabling AI agents to query and interact with multiple databases (BigQuery, ClickHouse, CockroachDB, Elasticsearch, etc.). Provides a standardized interface for LLM-based applications to access database functionality.

Tencent's WeKnora is an open-source LLM knowledge platform that converts raw documents into queryable RAG systems, autonomous reasoning agents, and self-maintaining wikis. Combines document processing, embeddings, and agentic AI in a single platform.

MCP server that feeds Figma design layouts to AI coding agents like Cursor, enabling AI to understand UI structure directly from design files. Bridges design-to-code workflow by giving agents visual context.

Memori is a memory infrastructure layer for AI agents that captures execution and conversation history into structured, persistent state. It's model-agnostic and designed for production systems.

Botpress is an open-source framework for building and deploying LLM agents with 14.7k stars. It provides a hub for creating AI-powered chatbots and agents using GPT and other models.

RWKV is an RNN architecture that combines transformer-like performance with linear-time inference and constant memory usage. It enables parallelizable training like GPT while offering infinite context length, no KV-cache overhead, and built-in sentence embeddings.

Comprehensive Chinese-language AI guide covering models (DeepSeek, GPT, Claude, Gemini), prompting, RAG, agents, tools (Cursor, Copilot), and frameworks. Open-source learning resource with 14k+ stars; useful for developers seeking curated AI knowledge and tutorials.