- prompts.chat by f, 177k stars. Community library of shareable AI prompts that can be self-hosted
- dify by langgenius, 161k stars. Platform for agentic workflows and RAG pipelines with model and tool support
- langchain by langchain-ai, 150k stars. Platform for building LLM agents and applications in Python
- awesome-mcp-servers by punkpeye, 98k stars. Curated list of MCP servers
- ragflow by infiniflow, 94k stars. Open source RAG engine with agent-oriented retrieval and document parsing
- servers by modelcontextprotocol, 92k stars. Reference implementations of MCP servers for common tools and services
- OmniRoute by cheaper-inference, 88k stars. Single endpoint gateway routing requests to hundreds of LLM providers
- crawl4ai by unclecode, 87k stars. Apache-2.0 web crawler that outputs markdown for LLM pipelines
- headroom by headroomlabs-ai, 80k stars. Compresses tool outputs and RAG chunks before they reach the LLM as a library, proxy, or MCP server
- openai-cookbook by openai, 77k stars. Example code and guides for calling the OpenAI API
- system_prompts_leaks by asgeirtj, 71k stars. Collected system prompts from major AI chatbots and coding tools
- free-claude-code by Alishahryar1, 70k stars. Proxy that runs Claude Code and similar coding harnesses on free model backends
- context7 by upstash, 65k stars. Hosted up-to-date library documentation served to AI agents over MCP
- litellm by BerriAI, 63k stars. Gateway and Python SDK that calls 100+ LLM APIs in OpenAI format
- CLIProxyAPI by router-for-me, 59k stars. Wraps AI coding CLIs as an OpenAI, Gemini and Claude compatible API
- chrome-devtools-mcp by ChromeDevTools, 56k stars. MCP server that gives coding agents access to Chrome DevTools
- claude-cookbooks by anthropics, 55k stars. Notebooks with recipes for using Claude's API
- freellmapi by tashfeenahmed, 52k stars. OpenAI-compatible gateway that routes requests across free LLM providers with failover
- new-api by QuantumNous, 52k stars. Self-hosted LLM gateway with OpenAI and Claude compatible APIs and usage logs
- JeecgBoot by jeecgboot, 48k stars. Low-code platform on Spring Boot and Vue with AI code generation
- iFixAi by ifixai-ai, 46k stars. Tool that audits and tests whether AI agents do what they should
- Fabric by danielmiessler, 45k stars. CLI that runs reusable AI prompt patterns from the command line
- faiss by facebookresearch, 42k stars. Library for fast nearest-neighbor search over dense vectors
- PageIndex by VectifyAI, 40k stars. Reasoning-based document index for retrieval without vector embeddings
- dspy by stanfordnlp, 39k stars. Framework for building LLM pipelines by writing code instead of prompts
- 1Panel by 1Panel-dev, 37k stars. Web control panel for Linux servers, Docker and AI model deployments
- WeKnora by Tencent, 37k stars. LLM knowledge platform that turns documents into RAG, an agent and a wiki
- 9router by decolua, 32k stars. Local gateway that routes coding tool requests across many AI providers
- openai-python by openai, 32k stars. Official Python client library for calling the OpenAI API
- fastmcp by PrefectHQ, 28k stars. Python framework for building MCP servers and clients
- ai by vercel, 28k stars. TypeScript SDK for building AI apps and agents with many model providers
- claude-quickstarts by anthropics, 21k stars. Starter projects for building deployable apps on the Claude API
- deepeval by confident-ai, 19k stars. Python framework with metrics for evaluating LLM outputs
- opencodex by lidge-jun, 19k stars. Proxy that routes Codex and Claude Code requests to other model providers
- jailbreaks by togg53192-cmd, 17k stars. LIST OF ALL MY JAILBREAKS
- experiential by experientiallabs, 17k stars. Gateway for BYOK and self-hosted models that routes traffic to cut cost
- zvec by alibaba, 16k stars. Embedded in-process vector database with HNSW indexing
- outlines by dottxt-ai, 16k stars. Library that constrains LLM output to a regex, JSON schema or grammar
- prompt-master by nidhinjs, 16k stars. Claude skill that writes prompts tailored to other AI tools
- codex-chatgpt-web by miuuyy, 16k stars. Lets Codex use the ChatGPT web interface as a model, with tools and images
New repos every two days, stars refreshed daily. Visits are counted without cookies, and the numbers are public. How it works Data RSS Newsletter