LightningRAG
A full-stack starter built with Vue and Gin that ships built-in retrieval-augmented generation: knowledge bases, vector search, agent orchestration on a visual canvas, and many LLM and vector-store integrations.
LightningRAG is a decoupled Vue frontend and Gin backend starter kit centered on retrieval-augmented generation. It ingests, parses and chunks documents into pluggable vector stores, supports hybrid and multi-path retrieval, and offers streaming chat with optional source references, backed by pluggable LLM, embedding and reranker providers.
An agent-orchestration canvas lets a developer wire retrieval, LLM calls, tools and control-flow nodes into multi-step or branching automations, with an extensible tool registry. Optional webhook connectors bind published agents to Feishu, DingTalk, WeChat, Slack, Teams and other chat platforms. The starter also includes JWT authentication, Casbin-based authorization, dynamic routes and menus, a form builder, and code generation, and is written in Go.