Open source AI tools

Open source AI tools: products with their code in a public repository under an open licence, most starred first, each with its licence and what it does.

The 300 most starred of 2925 open source AI tools

OpenClaw — Self-hosted personal agent with no subscription, run by a nonprofit (Agents, MIT, 393.5k stars on GitHub)
Superpowers — It enforces structured, test-driven development workflows for coding agents. (Code, MIT, 310.1k stars on GitHub)
ECC — It provides a shared operating layer and security scanning across multiple coding agents. (Code, MIT, 294.1k stars on GitHub)
DeepSeek Harness — Builds on a plugin architecture to organize documents, analyze tables, and write code. (Code, MIT, 268.5k stars on GitHub)
Hermes Agent — It features persistent memory, self-created skills, and isolated sandboxing. (Agents, MIT, 259.5k stars on GitHub)
OpenCode — Works with 75+ model providers, local models, and existing subscriptions (Code, MIT, 223.6k stars on GitHub)
Firecrawl — Provides direct web data access for AI agents with open source flexibility. (Search, AGPL-3.0, 197.7k stars on GitHub)
MarkItDown — Converts documents into LLM-friendly Markdown for better token efficiency. (Productivity, MIT, 193.4k stars on GitHub)
Ollama — Run open models locally or in the cloud with private prompts (Chat, MIT, 185.6k stars on GitHub)
Ponytail — Cuts unneeded code and dependencies by enforcing a minimal-code ladder. (Code, MIT, 181.5k stars on GitHub)
prompts.chat — Free, open-source community for sharing and organizing AI prompts. (Productivity, Other, 175.5k stars on GitHub)
Hugging Face Transformers — It provides a unified API for over 1M+ pretrained model checkpoints on the Hub. (Code, Apache-2.0, 168.4k stars on GitHub)
Dify — Open-source visual builder for agents and RAG apps (Agents, Other, 160.1k stars on GitHub)
Langflow — Builds and deploys AI agents visually with Python and major LLMs. (Agents, MIT, 156.2k stars on GitHub)
CC Switch — Centralizes AI coding tool management, API routing, and usage tracking in one interface. (Code, MIT, 150.9k stars on GitHub)
Spec Kit — Provides structured processes and templates for AI coding agents. (Code, MIT, 145k stars on GitHub)
ComfyUI — Local image and video generation on a node canvas, with every model and parameter in reach (Image, GPL-3.0, 144.5k stars on GitHub)
Graphify — It replaces grepping with traceable graph paths, reducing token usage significantly. (Code, Apache-2.0, 140.6k stars on GitHub)
gstack — It provides twenty-three specialist roles and eight power tools for Claude Code. (Code, MIT, 138.6k stars on GitHub)
MoneyPrinterTurbo — Generates complete short videos with scripts, voiceovers, and captions in one click. (Video, MIT, 137.2k stars on GitHub)
Browser Use — Managed remote browsers built for protected sites at $0.02 an hour. (Agents, MIT, 119.7k stars on GitHub)
TradingAgents — It uses specialized AI agents to collaboratively evaluate market conditions. (Agents, Apache-2.0, 119.5k stars on GitHub)
Caveman — Cuts 65 percent of tokens by communicating in concise caveman style. (Code, Apache-2.0, 116.5k stars on GitHub)
Orca — Runs multiple AI coding agents simultaneously in isolated workspaces. (Code, MIT, 112.7k stars on GitHub)
Agent Skills — Encodes senior-engineer workflows and quality gates into 24 skills. (Code, MIT, 112.5k stars on GitHub)
claude-mem — It shares persistent, searchable agent memory across sessions and devices. (Code, Apache-2.0, 109.5k stars on GitHub)
Paperclip — Orchestrate AI agents with org charts, budgets, and governance in one app. (Agents, MIT, 109.5k stars on GitHub)
Taste Skill — Prevents AI from generating generic, templated frontend code. (Code, MIT, 107.9k stars on GitHub)
Agent Reach — Provides zero API fees, multi-backend routing, and local credential storage. (Agents, MIT, 107.6k stars on GitHub)
HTML Anything — Reuses your local coding-agent CLI without requiring new API keys or accounts. (Image, Apache-2.0, 105.3k stars on GitHub)
Impeccable — It detects and removes default AI visual traits and respects your design system. (Code, Apache-2.0, 99.5k stars on GitHub)
VoiceStudio — Runs locally on your machine with no account, cloud connection, or API keys required. (Audio, AGPL-3.0, 98k stars on GitHub)
vLLM — Maximizes throughput with PagedAttention and continuous batching. (Code, Apache-2.0, 96k stars on GitHub)
vLLM Ascend — It enables vLLM functionality on Huawei Ascend NPU hardware. (Code, Apache-2.0, 95.7k stars on GitHub)
RAGFlow — Provides an integrated platform combining data processing with agent workflows. (Research, Apache-2.0, 93.3k stars on GitHub)
Zed — Very fast native editor with parallel agents and multiplayer (Code, Other, 92.4k stars on GitHub)
OpenHands — Runs autonomous agents locally, in the cloud, or inside your VPC with complete control. (Code, MIT, 92.4k stars on GitHub)
Archify — Generates interactive HTML diagrams with motion and export from code or descriptions. (Code, Apache-2.0, 92.4k stars on GitHub)
PaddleOCR — Open-source OCR and document parsing for 100+ languages (Productivity, Apache-2.0, 92k stars on GitHub)
Scrapling — It uses adaptive element tracking to maintain scrapers when websites change. (Code, BSD-3-Clause, 91.8k stars on GitHub)
Memory MCP server — Simple local memory that carries facts across chats (Code, Other, 91.6k stars on GitHub)
Sequential Thinking MCP server — Structured step-by-step reasoning any MCP client can call (Code, Other, 91.6k stars on GitHub)
World Monitor — It correlates live geopolitical, shipping, and market signals on a single map. (Research, AGPL-3.0, 89.9k stars on GitHub)
Understand Anything — Turns complex code into meaningful business domain flows and interactive stories. (Code, MIT, 89.6k stars on GitHub)
Crawl4AI — Turns any URL into clean Markdown or typed JSON for AI models. (Search, Apache-2.0, 86.7k stars on GitHub)
RTK — Reduces AI context output noise by 56 to 90 percent with zero configuration. (Code, Apache-2.0, 85.2k stars on GitHub)
DeerFlow — Executes long tasks using sandboxes, memories, tools, and subagents. (Agents, MIT, 84.8k stars on GitHub)
OmniRoute — Routes requests across 352 providers with automatic fallback and cost-based selection. (Code, MIT, 83.6k stars on GitHub)
Ruflo Console — Manages architecture decision records, swarms, and sessions inside Claude Code. (Agents, MIT, 79.9k stars on GitHub)
Career-Ops — Runs locally in your CLI with open-source code and zero telemetry. (Agents, MIT, 79.3k stars on GitHub)
Unsloth — Runs and trains models locally with lower memory usage. (Code, Apache-2.0, 78.9k stars on GitHub)
Headroom — Reduces token usage by 60 to 95 percent for JSON and lowers costs while maintaining the same answers. (Code, Apache-2.0, 78.5k stars on GitHub)
AppFlowy — Runs on your infrastructure and keeps AI processing inside your environment. (Productivity, AGPL-3.0, 78k stars on GitHub)
HyperFrames — It enables AI agents to create and render production-ready videos. (Video, Apache-2.0, 76.6k stars on GitHub)
CodeGraph — It eliminates file crawling, reducing AI agent tool calls by 88% and cutting costs. (Code, MIT, 76.1k stars on GitHub)
LlamaFactory — Provides a code-free CLI and Web UI for training 100+ different LLMs. (Code, Apache-2.0, 75.8k stars on GitHub)
OpenMontage — Turns your AI coding assistant into a full video production studio. (Video, AGPL-3.0, 74.4k stars on GitHub)
Strix — Continuous autonomous pentesting with auto-fix pull requests. (Code, Apache-2.0, 73.2k stars on GitHub)
PPT Master — Generates real PowerPoint decks with native shapes, charts, and slide masters instead of flat images. (Productivity, MIT, 72.1k stars on GitHub)
Cline — Open source, any model, with approval on every step (Code, Apache-2.0, 71.1k stars on GitHub)
Docling — Provides advanced PDF understanding and local execution for sensitive data. (Productivity, MIT, 69.9k stars on GitHub)
AI Job Search — Automates the full job application lifecycle using a local, private workflow. (Agents, MIT, 68.7k stars on GitHub)
Mem0 — It provides drop-in persistent memory for AI agents without changing pipelines. (Agents, Apache-2.0, 68.2k stars on GitHub)
AnythingLLM — Runs fully on-device with no accounts, API keys, or token limits required. (Chat, MIT, 67.8k stars on GitHub)
Daily Stock Analysis — Integrates multi-source data, LLM decisions, and multi-channel notifications. (Research, MIT, 67.4k stars on GitHub)
Last30Days Skill — Searches Reddit, X, YouTube, and Polymarket in parallel and scores by engagement. (Research, MIT, 66.2k stars on GitHub)
Warp — AI terminal plus cloud agents with built-in evals and benchmarks (Code, AGPL-3.0, 65.8k stars on GitHub)
GPT-SoVITS — It allows high-fidelity voice cloning with as little as one minute of data. (Audio, MIT, 64.5k stars on GitHub)
Context7 — Provides up-to-date documentation for AI agents and code editors. (Code, MIT, 64k stars on GitHub)
Diagram Design — Generates editorial-quality visual types matched to brand colors and typography in seconds. (Code, MIT, 64k stars on GitHub)
Humanizer — Integrates directly into agent workflows to rewrite text. (Writing, MIT, 63.5k stars on GitHub)
TrendRadar — Provides automated multi-platform tracking with natural language AI filtering. (Research, GPL-3.0, 63.3k stars on GitHub)
Ultralytics YOLO — Delivers high-speed and high-accuracy results in real-time. (Image, AGPL-3.0, 63k stars on GitHub)
LiteLLM — One OpenAI-format API for 100+ LLM providers (Code, Other, 62.5k stars on GitHub)
Plane — Assigns work to AI agents and answers questions using workspace context. (Agents, AGPL-3.0, 61.7k stars on GitHub)
CrewAI — Open-source agent framework with enterprise governance on top (Agents, MIT, 60.3k stars on GitHub)
MemPalace — Achieves high recall with local-first storage and zero API calls. (Agents, MIT, 60.2k stars on GitHub)
Meilisearch — It returns answers in less than 50 milliseconds with out-of-the-box relevancy. (Search, Other, 59.7k stars on GitHub)
I Have ADHD — Stops coding agents from burying answers with direct, action-first steps. (Code, MIT, 59.6k stars on GitHub)
Voicebox — Perform voice cloning and generation entirely on local hardware. (Audio, MIT, 59.2k stars on GitHub)
Marketing Skills — Provides 50 specific marketing skills for AI coding environments. (Agents, MIT, 57.8k stars on GitHub)
Appwrite — Build and scale apps rapidly with a unified platform for agents and developers. (Code, BSD-3-Clause, 57.8k stars on GitHub)
PrivateGPT — Popular open-source project ensuring 100% privacy for AI applications. (Chat, Apache-2.0, 57.6k stars on GitHub)
Hindsight — It uses a four-way retrieval strategy and background consolidation to maintain accurate memory. (Agents, MIT, 57k stars on GitHub)
Security Audit Skill — Automated multi-stage security auditing with independent verification. (Code, MIT, 56.5k stars on GitHub)
Goose — Runs locally on your machine with support for 70+ extensions and any LLM. (Agents, Apache-2.0, 56.1k stars on GitHub)
Scientific Agent Skills — Provides 177 validated skills and 100+ scientific databases for research agents. (Research, MIT, 55.6k stars on GitHub)
Flowise — Modular building blocks and visual tools to build and scale agentic systems fast. (Agents, Other, 55.5k stars on GitHub)
VibeVoice — Process long-form audio up to 60 minutes in a single pass. (Audio, MIT, 55k stars on GitHub)
whisper.cpp — It provides high-performance, offline speech recognition on diverse hardware. (Audio, MIT, 54.8k stars on GitHub)
Chrome DevTools MCP — Enables coding agents to interact directly with a real browser instance. (Code, Apache-2.0, 54.6k stars on GitHub)
Dsh Our Free Model — Provides free and unlimited access to frontier AI models with zero configuration. (Code, MIT, 54.5k stars on GitHub)
DeepTutor — It organizes learning into workspaces with custom AI agents and knowledge bases. (Agents, Apache-2.0, 53.7k stars on GitHub)
Cherry Studio — Unifies 300+ AI models and local backups in one open-source client. (Chat, AGPL-3.0, 53.1k stars on GitHub)
LlamaIndex — It provides hosted APIs for parsing documents and extracting typed data with one API key. (Code, MIT, 52.7k stars on GitHub)
New API — Unifies access to 30 plus AI providers with intelligent routing and cost control. (Code, AGPL-3.0, 51.1k stars on GitHub)
Shannon — Provides validated security findings by executing real exploits against your code. (Code, AGPL-3.0, 50.8k stars on GitHub)
Obsidian Skills — Integrates AI agents directly with Obsidian vaults and markdown files. (Productivity, MIT, 50.5k stars on GitHub)
Huginn — It acts as a self-hosted version of IFTTT or Zapier. (Agents, MIT, 50.2k stars on GitHub)
LocalAI — Runs text, voice, and vision workloads on ordinary hardware with an OpenAI-compatible API. (Code, MIT, 50.1k stars on GitHub)
Metabase — Query and visualize data from any database using plain English, SQL, or clicks. (Research, Other, 50k stars on GitHub)
Aider — Free terminal agent with a repo map and automatic git commits (Code, Apache-2.0, 49.8k stars on GitHub)
nanobot — Provides a compact, self-hosted agent runtime with long-horizon execution. (Agents, MIT, 49.7k stars on GitHub)
Tv Vibehack — Removes bloatware and replaces the stock launcher using ADB and an AI assistant. (Productivity, Other, 48k stars on GitHub)
Codebase Memory MCP — High-speed indexing and persistent knowledge graphs for AI coding agents. (Code, MIT, 47.9k stars on GitHub)
CowAgent — Open-source agent with multi-model and multi-channel support. (Agents, MIT, 47.7k stars on GitHub)
Text Generation WebUI — Run local LLMs with multiple backends, multimodal input, and tool calling 100% offline. (Chat, AGPL-3.0, 47.7k stars on GitHub)
Herdr — Persistent terminal sessions for coding agents that survive shutdowns. (Code, Apache-2.0, 47.3k stars on GitHub)
SiYuan — Privacy-first personal knowledge management with local storage and block-based editing. (Productivity, AGPL-3.0, 47k stars on GitHub)
LibreChat — Unifies multiple AI models and tools in one customizable, open-source platform. (Chat, MIT, 46.6k stars on GitHub)
Milvus — Distributed vector search that scales to billions of vectors (Code, Apache-2.0, 46.5k stars on GitHub)
FreeLLMAPI — Accesses hundreds of free LLM models and billions of tokens through one API key. (Code, MIT, 45.8k stars on GitHub)
Jan — Runs 100 percent offline on your computer. (Chat, Other, 45.2k stars on GitHub)
Hallmark — Generates distinct web designs and avoids common AI visual clichés. (Code, MIT, 45k stars on GitHub)
Fabric — It provides a modular system for solving specific problems using AI. (Productivity, MIT, 44.5k stars on GitHub)
Ray — It orchestrates distributed infrastructure for any AI workload at any scale. (Code, Apache-2.0, 44.1k stars on GitHub)
LangGraph — Fine-grained control over long-running, stateful agent workflows (Agents, MIT, 44k stars on GitHub)
Gradio — Builds web interfaces for machine learning models using only Python code. (Code, Apache-2.0, 43.9k stars on GitHub)
Reverse Skill — Routes AI agents to exact security methodologies instead of guessing commands. (Code, MIT, 43.9k stars on GitHub)
DeepSpeed — It supports training models with over 100 billion parameters. (Code, Apache-2.0, 43.2k stars on GitHub)
Agno — Build your agents once and serve them across multiple platforms. (Code, Apache-2.0, 43k stars on GitHub)
iFixAi — Audits agent behavior independently without building manual eval harnesses. (Agents, Apache-2.0, 42.5k stars on GitHub)
OpenHuman — Runs locally on your laptop with encrypted memory and native app integrations. (Agents, GPL-3.0, 42.4k stars on GitHub)
OpenMAIC — Build structured, multi-agent courses from documents in one click. (Agents, MIT, 42.4k stars on GitHub)
Chatbox — Chat with multiple AI models and local documents using your own API keys. (Chat, GPL-3.0, 42.2k stars on GitHub)
Project NOMAD — Provides access to large-scale knowledge bases without requiring an internet connection. (Research, Apache-2.0, 41.6k stars on GitHub)
AirLLM — Runs massive 70B+ parameter models on small 4GB consumer GPUs. (Code, Apache-2.0, 41.5k stars on GitHub)
BitNet — Runs 1-bit models on CPUs and GPUs with significantly reduced energy consumption. (Code, MIT, 41.4k stars on GitHub)
OpenViking — Unifies agent memory, knowledge RAG, and skills into a single directory structure. (Agents, AGPL-3.0, 41.3k stars on GitHub)
ToolJet — Builds complete apps and workflows using natural language or coding agents. (Code, AGPL-3.0, 41.1k stars on GitHub)
Codewhale — Runs in the terminal with open-source agents and local or hosted AI models. (Code, MIT, 41.1k stars on GitHub)
Appsmith — Connects to any database, SaaS tool, or API for rapid custom application development. (Code, Apache-2.0, 41.1k stars on GitHub)
Open Notebook — It maintains privacy while combining artificial intelligence with note-taking workflows. (Research, MIT, 40.7k stars on GitHub)
Oh My ClaudeCode — Orchestrates parallel Claude, Gemini, and Codex CLI workers. (Code, MIT, 40.4k stars on GitHub)
LightRAG — It uses a dual-layer architecture to reduce LLM calls and retrieval latency. (Research, MIT, 40.4k stars on GitHub)
Novu — Connects AI agents and products to customers across multiple channels with one API. (Code, Other, 40.2k stars on GitHub)
PageIndex — It provides 98.7% accuracy on FinanceBench without vector databases. (Research, MIT, 39.9k stars on GitHub)
MindsHub — Pay-as-you-go access to multiple AI models with local agent integration. (Agents, MIT, 39.8k stars on GitHub)
Quivr — It provides a pre-built RAG infrastructure for rapid application integration. (Code, MIT, 39.6k stars on GitHub)
LangExtract — It maps every extraction to its exact location in the source text. (Code, Apache-2.0, 39.2k stars on GitHub)
Playwright MCP — Enables LLMs to interact with web pages using structured accessibility snapshots. (Code, Apache-2.0, 38.9k stars on GitHub)
llmfit — Auto-detects hardware to recommend and score compatible open-source LLMs. (Code, MIT, 38.5k stars on GitHub)
CopilotKit — Connects your application to AI agents with generative UI and human approvals. (Code, MIT, 38.4k stars on GitHub)
OpenBot — Runs entirely on your own infrastructure with strict document permissions and custom. (Agents, MIT, 38.4k stars on GitHub)
SGLang — It delivers high-throughput, low-latency serving for large-scale deployments. (Code, Apache-2.0, 38.3k stars on GitHub)
PDFMathTranslate — Translates PDF files while preserving original layouts and math formulas. (Writing, AGPL-3.0, 38.1k stars on GitHub)
Vane — Runs entirely on your hardware with local and cloud LLMs. (Search, MIT, 37.9k stars on GitHub)
Claude Code Router — It centralizes control of AI agent requests with intelligent routing and observability. (Code, MIT, 37.9k stars on GitHub)
Khoj — Conducts deep research across personal documents and the web. (Research, AGPL-3.0, 37.9k stars on GitHub)
Oh My Pi — Runs natively on Windows without requiring WSL, powered by a fast Rust core. (Code, MIT, 37.4k stars on GitHub)
Vibe-Trading — Provides reproducible, inspectable research artifacts for financial markets. (Agents, MIT, 37.3k stars on GitHub)
Strata — Runs large AI models locally on consumer gaming PCs without server reliance. (Code, MIT, 37.1k stars on GitHub)
Langfuse — It connects tracing, prompt management, and evaluations in one open platform. (Code, Other, 36.8k stars on GitHub)
Handy — It runs completely offline and keeps your voice data private on your computer. (Audio, MIT, 36.6k stars on GitHub)
Buzz — Humans and agents share the exact same keys, rooms, and audit trails. (Agents, Apache-2.0, 36.6k stars on GitHub)
GraphRAG — Enhances LLM reasoning over private data using knowledge graph structures. (Research, MIT, 36.5k stars on GitHub)
AgentGPT — It allows users to build and deploy autonomous agents in the browser. (Agents, GPL-3.0, 36.2k stars on GitHub)
WeKnora — Combines RAG, multi-step agents, and automatic Wiki generation in one platform. (Research, Other, 36.2k stars on GitHub)
Qdrant — Fast filtered vector search, self-hosted or in a managed cloud (Code, Apache-2.0, 35.3k stars on GitHub)
SillyTavern — Local LLM chat UI with lorebooks, extras and many backends (Chat, AGPL-3.0, 35.2k stars on GitHub)
Diffusers — A modular toolbox for inference and training of diffusion models. (Image, Apache-2.0, 34.8k stars on GitHub)
Codex Plugin for Claude Code — Delegate tasks and run code reviews with Codex directly inside Claude Code. (Code, Apache-2.0, 34.7k stars on GitHub)
Laya — It runs typed decisions in a single forward pass over 100+ languages. (Research, Apache-2.0, 34.7k stars on GitHub)
ChatDev — Build and execute customized multi-agent systems rapidly through simple configuration without coding. (Code, Apache-2.0, 34.6k stars on GitHub)
Knowledge Work Plugins — Customize Claude with company-specific tools, terminology, and processes. (Productivity, Apache-2.0, 34.3k stars on GitHub)
OpenClaude — Connects one terminal interface to over 200 different AI models. (Code, Other, 34.2k stars on GitHub)
GitHub MCP Server — Connects AI tools directly to GitHub for natural language code management. (Code, MIT, 34.1k stars on GitHub)
Tabby — Runs self-hosted with local-first deployment and open-source transparency. (Code, Other, 33.9k stars on GitHub)
Context Mode — Saves up to 98 percent of context per session by keeping raw data out of the LLM. (Code, Other, 33.1k stars on GitHub)
Onyx — Secures sensitive data by running entirely in your own cloud or on-premise infrastructure. (Search, Other, 33.1k stars on GitHub)
ZeroClaw — Own your AI infrastructure with no subscriptions or cloud dependencies. (Agents, Apache-2.0, 33k stars on GitHub)
MCP for Blender — It allows direct control of Blender 3D via LLM integration. (Image, MIT, 33k stars on GitHub)
Cognee — Persistent memory for AI agents using graph, vector, and relational retrieval. (Agents, Apache-2.0, 32.9k stars on GitHub)
OfficeCLI — Single binary, open-source, and gives AI agents visual feedback without needing Office installed. (Productivity, Apache-2.0, 32.9k stars on GitHub)
Code Review Graph — Provides a median 63x context reduction while keeping all code processing local. (Code, MIT, 32.5k stars on GitHub)
SigNoz — Unifies logs, metrics, and traces into one correlated platform. (Code, Other, 32.5k stars on GitHub)
Claude Code Game Studios — It brings structured studio hierarchy and automated workflows to solo AI game development. (Code, MIT, 32.4k stars on GitHub)
ScrapeGraphAI — Extracts structured data via natural language prompts with no selectors or proxies. (Search, MIT, 32.4k stars on GitHub)
Graphiti — Provides incremental updates and hybrid retrieval for knowledge graphs. (Agents, Apache-2.0, 32.4k stars on GitHub)
Meetily — It keeps all transcription and audio processing 100% local on your device. (Productivity, MIT, 32.3k stars on GitHub)
dots — Uses a patched Firefox engine to bypass detection and web blocks. (Agents, MIT, 32.3k stars on GitHub)
Google Workspace CLI — Dynamically builds commands from Google's Discovery Service with JSON output. (Productivity, Apache-2.0, 32k stars on GitHub)
Frontend Slides — Generates stunning HTML presentations with visual style previews and no dependencies. (Productivity, MIT, 31.7k stars on GitHub)
Cua — Scales computer fleets across four OS families with one MCP/CLI driver. (Agents, MIT, 31.5k stars on GitHub)
GBrain — Shares team memory across multiple AI tools with one flat workspace price. (Agents, MIT, 31.4k stars on GitHub)
NanoClaw — Runs in isolated Docker containers with credential protection via OneCLI. (Agents, MIT, 31k stars on GitHub)
Serena — It offers symbol-level code retrieval and refactoring tools for AI agents. (Code, Other, 30.9k stars on GitHub)
Composio — Equips AI agents with tools, managed user auth, and sandboxes across 1,500+ apps (Agents, MIT, 30.8k stars on GitHub)
OpenAI Agents SDK — Provider-agnostic multi-agent framework with built-in guardrails and tracing. (Code, MIT, 30.4k stars on GitHub)
GPT Researcher — Ranked #1 on Carnegie Mellon's DeepResearchGym benchmark for academic research. (Research, Apache-2.0, 30.4k stars on GitHub)
Video Use — Edits raw footage using coding agents with word-boundary precision and self-evaluation. (Video, MIT, 30.3k stars on GitHub)
Karakeep — Provides AI-based automatic tagging and semantic search for saved content. (Productivity, AGPL-3.0, 30.2k stars on GitHub)
OpenDataLoader PDF — It ranks first in benchmarks for layout and table extraction without needing a GPU. (Productivity, Apache-2.0, 30.1k stars on GitHub)
AgentMemory — It runs locally with zero external databases and achieves high retrieval recall. (Code, Apache-2.0, 30k stars on GitHub)
Obscura — Boots clean, isolated browser sessions in under 50 milliseconds. (Agents, Apache-2.0, 29.9k stars on GitHub)
FastGPT — It uses hybrid retrieval and auto-cleansed data to prevent hallucinations. (Agents, Other, 29.9k stars on GitHub)
Sim — Centralized governance, over 1,000 integrations, and self-hosting options. (Agents, Apache-2.0, 29.9k stars on GitHub)
Pixelle-Video — It uses a modular design to combine LLMs, TTS, and video models via ComfyUI or direct API connections. (Video, Apache-2.0, 29.6k stars on GitHub)
Chroma — Object storage architecture with automatic query-aware data tiering and caching. (Code, Apache-2.0, 29.6k stars on GitHub)
Page Agent — Integrates directly into web pages without browser extensions or headless browsers. (Agents, MIT, 29.6k stars on GitHub)
Nx — It caches every task result and only reruns what your code changes touched. (Code, MIT, 29.4k stars on GitHub)
Pydantic AI — Provides type safety, structured outputs, validation, and durable execution for reliable agents. (Code, MIT, 29.2k stars on GitHub)
Mastra — Provides agents, workflows, memory, and observability in one TypeScript framework. (Code, Other, 29.2k stars on GitHub)
Claude Skills Pack — 388 production-ready skills across 13 tools with zero-dependency Python tools. (Code, MIT, 29.1k stars on GitHub)
Repomix — Packages entire codebases into single files for AI context. (Code, MIT, 29.1k stars on GitHub)
Guizang PPT Skill — Generates structured, professional HTML decks directly inside AI agent workflows. (Productivity, AGPL-3.0, 29k stars on GitHub)
Claude HUD — Shows real-time context usage, rate limits, and agent activity inside your terminal. (Code, MIT, 29k stars on GitHub)
Resume Matcher — Compare resumes against job descriptions to pass automated tracking systems. (Writing, Apache-2.0, 29k stars on GitHub)
Invoke — Offers an industry-leading web UI and a fully integrated unified canvas. (Image, Apache-2.0, 28.9k stars on GitHub)
Qwen Code — Fully open-source framework and models with multi-protocol support. (Code, Apache-2.0, 28.8k stars on GitHub)
cmux — It features vertical tabs, notification rings, and split panes for AI coding. (Code, Other, 28.6k stars on GitHub)
MLflow — Complete observability and evaluation for production AI apps. (Code, Apache-2.0, 28.6k stars on GitHub)
Label Studio — Supports all data modalities and custom layouts for diverse AI pipelines. (Code, Apache-2.0, 28.6k stars on GitHub)
Vibe Kanban — Run parallel coding agents and automate git worktrees for faster reviews. (Code, Apache-2.0, 28.5k stars on GitHub)
DBX — It combines lightweight size, open-source access, and built-in AI for over one hundred databases. (Code, Apache-2.0, 28.4k stars on GitHub)
Huashu Design — Eliminates the need for manual design tools by turning natural language into production-ready assets. (Code, MIT, 28.4k stars on GitHub)
Beads — Prevents context loss with version-controlled, dependency-aware storage. (Code, MIT, 28.4k stars on GitHub)
Budibase — Builds AI agents and automations using your own data and LLMs. (Agents, Other, 28.3k stars on GitHub)
Chat2DB — Local-first data processing keeps your database schema private and secure. (Code, Other, 28.3k stars on GitHub)
FastMCP — The standard framework for building Model Context Protocol servers and clients. (Code, Apache-2.0, 28.2k stars on GitHub)
Kilo Code — Open-source agent with any model and no markup on inference (Code, MIT, 27.8k stars on GitHub)
Magpie — It enables AI agents to use models not natively supported by their internal protocols. (Code, MIT, 27.7k stars on GitHub)
Grok Build — Runs parallel subagents in git worktrees directly from your terminal (Code, Apache-2.0, 27.6k stars on GitHub)
PM Skills — Provides structured frameworks and judgment for AI rather than generic text. (Productivity, MIT, 27.6k stars on GitHub)
AgenticSeek — It runs entirely on a local machine without cloud dependency or API costs. (Agents, GPL-3.0, 27.5k stars on GitHub)
Hypit — Clone viral videos and generate 100 variants with a single command. (Video, Other, 27.3k stars on GitHub)
Baoyu Skills — Improves daily work efficiency with AI agents using ready-to-use skills. (Productivity, MIT, 27.1k stars on GitHub)
Onlook — Edits live React components using the real Storybook design system. (Code, Apache-2.0, 27k stars on GitHub)
Haystack — Modular building blocks provide full visibility to inspect, debug, and optimize decisions. (Code, Apache-2.0, 26.9k stars on GitHub)
Agent Skills Spec — Capture specialized knowledge as reusable instructions for AI agents. (Agents, Apache-2.0, 26.8k stars on GitHub)
Promptfoo — It simulates real users to uncover agent and RAG vulnerabilities. (Code, MIT, 26.7k stars on GitHub)
faster-whisper — Transcribes audio faster with lower memory usage than standard implementations. (Audio, MIT, 26.3k stars on GitHub)
Crawlee — Handles blocking, crawling, proxies, and browsers for JavaScript and Python. (Search, Apache-2.0, 26.3k stars on GitHub)
Distilly — It distills real human experience and judgment into portable agent skills. (Agents, MIT, 26.2k stars on GitHub)
Flux — Runs via API or as open weights on your own infrastructure. (Image, Apache-2.0, 26k stars on GitHub)
Claude SEO — Free open-source SEO toolkit running directly inside the terminal. (Agents, MIT, 26k stars on GitHub)
OpenSEO — Usage-based billing and open source architecture reduce SEO tool costs. (Agents, MIT, 26k stars on GitHub)
Watermarks Remover — Removes invisible AI provenance marks and metadata from your files. (Image, MIT, 25.9k stars on GitHub)
Stagehand — It combines familiar APIs with self-healing actions and token efficiency for agents. (Agents, MIT, 25.8k stars on GitHub)
Kotaemon — It offers an open-source, customizable interface for local and API-based document retrieval. (Research, Apache-2.0, 25.8k stars on GitHub)
Readest — Read EPUB and PDF files with built-in translation, text-to-speech, and cross-device sync. (Productivity, AGPL-3.0, 25.8k stars on GitHub)
TOON — It reduces token usage by 42.6% while maintaining JSON-equivalent accuracy. (Code, MIT, 25.5k stars on GitHub)
Letta Code — It features persistent memory, sleep-time compute, and git-tracked context. (Code, Apache-2.0, 25.3k stars on GitHub)
Activepieces — Unifies AI agents, chat, flows, tables, and 760+ apps with open-source core and self-hosting options. (Agents, Other, 25.3k stars on GitHub)
AGENTS.md — It separates agent-specific instructions from human-focused README files. (Code, MIT, 25.2k stars on GitHub)
NextAI Translator — Supports translation, polishing, and summarization across 55 languages. (Writing, AGPL-3.0, 25k stars on GitHub)
WhisperX — It enables 70x faster, batched transcription with high-accuracy alignment. (Audio, BSD-2-Clause, 24.9k stars on GitHub)
IndexTTS — Supports zero-shot voice cloning with multi-language emotion and speed control. (Audio, Other, 24.8k stars on GitHub)
Text-to-CAD — Generates CAD models, drawings, and robot files from plain text or images. (Agents, MIT, 24.8k stars on GitHub)
Happy — Control coding agents remotely with end-to-end encryption. (Code, MIT, 24.4k stars on GitHub)
RAG-Anything — It unifies text and multimodal document retrieval within a single framework. (Research, MIT, 24.3k stars on GitHub)
SuperClaude — Adds 30 slash commands and 8 MCP servers to Claude Code. (Code, MIT, 23.9k stars on GitHub)
Anatomy — Builds interactive isometric technical illustrations with precise parts and live data. (Code, MIT, 23.8k stars on GitHub)
Vanna AI — It scales from prototype to multi-tenant production without rewrites. (Research, MIT, 23.8k stars on GitHub)
Archon — Orchestrate multiple AI agents with isolated workflows via a single command layer. (Code, MIT, 23.7k stars on GitHub)
AudioCraft — Single codebase for generative music, sound effects, and audio compression. (Audio, MIT, 23.7k stars on GitHub)
Skyvern — Vision-based browser automation that survives layout changes (Agents, AGPL-3.0, 23.3k stars on GitHub)
MLC LLM — Enables native optimization and deployment of AI models across platforms. (Code, Apache-2.0, 23.2k stars on GitHub)
oMLX — It uses paged SSD KV caching to reduce agent TTFT to under 5 seconds. (Code, Apache-2.0, 23.2k stars on GitHub)
n8n-MCP — Connects AI tools directly to n8n for zero-setup workflow creation and debugging. (Agents, MIT, 23.1k stars on GitHub)
Opik — It offers comprehensive tracing, automated evaluations, and open-source flexibility. (Code, Apache-2.0, 23.1k stars on GitHub)
MaxKB — Open-source, self-hosted, and production-proven enterprise agent platform. (Agents, GPL-3.0, 23k stars on GitHub)
CodexBar — See Codex, Claude, Cursor and 86 other limits in the menu bar (Code, MIT, 22.9k stars on GitHub)
NemoClaw — Provides managed security, networking, and lifecycle operations for AI agents. (Agents, Apache-2.0, 22.9k stars on GitHub)
Paseo — Run coding agents on your machine and connect from phone, desktop, or web. (Code, Other, 22.8k stars on GitHub)
Univer — Integrates high-performance office tools and AI agents into your application. (Productivity, Apache-2.0, 22.8k stars on GitHub)
OpenObserve — Full observability stack at a fraction of the usual storage cost (Agents, AGPL-3.0, 22.6k stars on GitHub)
Wave Terminal — Integrates file previews, editing, AI, and web browsing directly into your terminal. (Code, Apache-2.0, 22.6k stars on GitHub)
Prime Agent — It combines a persistent Python environment with durable state for long tasks. (Code, Other, 22.5k stars on GitHub)
Buzz Transcriber — Transcribes and translates audio offline directly on personal computers. (Audio, MIT, 22.3k stars on GitHub)
Airbyte — Connects 700+ sources to warehouses, lakes, and AI agents with sovereign hybrid deployment. (Code, Other, 22.2k stars on GitHub)
LocalGPT — Runs fully on-premise to keep all processed data completely private. (Chat, MIT, 22.1k stars on GitHub)
Docmost — It offers secure, self-hosted deployment for strict data compliance requirements. (Productivity, AGPL-3.0, 22.1k stars on GitHub)
Dyad — Runs locally on your machine with zero cloud lock-in and support for any AI model. (Code, Other, 22.1k stars on GitHub)
Harness2 — It recursively refines agent execution structures using parallel and sequential search. (Agents, Apache-2.0, 22.1k stars on GitHub)
Google ADK — It applies software development principles to agent creation with Python. (Code, Apache-2.0, 22k stars on GitHub)
LiveKit Agents — Connects AI models to WebRTC rooms with reliable turn detection and streaming pipelines. (Audio, Apache-2.0, 22k stars on GitHub)
PEFT — Reduces computational requirements and storage needs for large model training. (Code, Apache-2.0, 21.8k stars on GitHub)
Humanizer ZH — Removes AI phrasing from Chinese text while preserving facts and author voice. (Writing, MIT, 21.7k stars on GitHub)
Khazix Skills — Practical open source AI skills used daily and proven in real projects. (Productivity, MIT, 21.7k stars on GitHub)
Coze Studio — Visual no-code or low-code tools for complete AI agent development and deployment. (Agents, Apache-2.0, 21.7k stars on GitHub)
Google Agent Skills — Access pre-configured automation skills for Google Cloud and developer tools. (Code, Apache-2.0, 21.7k stars on GitHub)
FinGPT — Access open-source financial language models and APIs at a low cost. (Research, MIT, 21.5k stars on GitHub)
jcode — Optimized for massive parallelism with low memory usage per session. (Code, MIT, 21.4k stars on GitHub)
Rasa — Provides building blocks to create virtual assistants and chatbots. (Agents, Apache-2.0, 21.3k stars on GitHub)
Ego Lite — Executes browser automation in isolated workspaces using existing logins. (Agents, MIT, 21.2k stars on GitHub)
FunASR — It supports offline, streaming, and edge deployment with multiple models. (Audio, MIT, 20.9k stars on GitHub)
Hummingbot — Runs agentic strategies on any CEX and DEX using open source Python code. (Agents, Apache-2.0, 20.9k stars on GitHub)
Voyager — Organize AI chats with folders and save prompts. (Chat, GPL-3.0, 20.5k stars on GitHub)
ebook2audiobook — Convert multiple ebook formats into audiobooks with voice cloning and 1158 languages. (Audio, Apache-2.0, 20.4k stars on GitHub)
DB-GPT — Builds autonomous agents for complex data tasks with private LLM technology. (Research, MIT, 20.2k stars on GitHub)
daily.dev — Access a ranked, ad-supported feed of developer news without paywalls. (Research, AGPL-3.0, 20.1k stars on GitHub)
notebooklm-py — Automates research, content generation, and structured data export from notebooks. (Research, MIT, 20k stars on GitHub)
img2threejs — Generates code-only, animation-ready Three.js models from a single image. (Image, Apache-2.0, 20k stars on GitHub)
Brag — Generates complete launch videos directly from your local project files. (Video, MIT, 19.9k stars on GitHub)
PDF Inspector — Detects if a PDF is scanned or text-based to avoid unnecessary OCR costs. (Productivity, MIT, 19.8k stars on GitHub)
Magenta — Builds and plays AI musical instruments with open models and real-time neural synthesis. (Audio, Apache-2.0, 19.7k stars on GitHub)
Agent Zero — It provides a transparent, controllable environment for autonomous AI agents. (Agents, Other, 19.6k stars on GitHub)
pyVideoTrans — Completely open source, free to use, and runs locally with no hardware requirements. (Video, GPL-3.0, 19.6k stars on GitHub)
RocketRide — It runs locally in your IDE with a C++ engine and 100+ pipeline nodes. (Code, MIT, 19.5k stars on GitHub)