OpenViking is an open-source context database for AI agents. It stores memories, resources, and skills as one virtual filesystem under the viking:// protocol, so an agent browses its own context with ls, tree, and find instead of querying a black-box vector store.
> Prefer to Clone Locally? > > This repository includes 50+ language translations which significantly increases the download size. To clone without translations, use sparse checkout: > > Bash / macOS / Linux: >
A community-driven hub of 42+ runnable notebooks covering RAG techniques from foundational to cutting-edge - the intuition, the code, and the references to build more accurate, context-rich retrieval systems.
Open-source AI orchestration framework for building context-engineered, production-ready LLM applications. Design modular pipelines and agent workflows with explicit control over retrieval, routing, memory, and generation.
企业级 Agentic RAG 智能体 - 全链路覆盖文档解析、多路检索、意图识别、问题重写、会话记忆、MCP 工具调用与深度思考。面向真实业务场景，从 0 到 1 完整工程实现。
STORM is a LLM system that writes Wikipedia-like articles from scratch based on Internet search. Co-STORM further enhanced its feature by enabling human to collaborative LLM system to support more aligned and preferred information seeking and knowledge curation.
Complete API layer for private AI applications on local models: RAG, skills, tools, MCP, text-to-sql, and more. Works with any OpenAI-compatible inference server.
A modular Agentic RAG built with LangGraph — learn Retrieval-Augmented Generation Agents in minutes.
Code search MCP for Claude Code. Make entire codebase the context for any coding agent.