登录

autoflow

pingcapApache-2.0活跃维护

基于 Graph RAG 的对话式知识库工具，用 TiDB Serverless 向量存储做检索，能基于知识图谱回答提问。

pingcap/autoflow is a Graph RAG based and conversational knowledge base tool built with TiDB Serverless Vector Storage. Demo: https://tidb.ai

0.4.0发布于 2025-01-03 09:33 UTC2,982star193fork同步于 2026-11-10 11:42 UTC
知识库图检索增强向量存储对话问答TiDB
对比 GitHub 官网chatbotcotgraphragknowledge-graphmysql

怎么上手

由 README 提炼，执行前请确认
原文 
docker compose up -d
硬件要求按 README 说明

README 未说明 GPU 或显存要求；Docker Compose 部署建议 4 核 CPU、8GB 内存。

有 Docker 镜像
下载：源码 zip·GitHub Releases（安装包）本站不镜像安装包

项目说明

由 README 自动翻译提炼
没人工点评？写下你的经验

基于 Graph RAG 的对话式知识库工具，用 TiDB 向量存储快速搭建站点问答

基于 Graph RAG 的对话式知识库工具，用 TiDB Serverless 向量存储做检索，能基于知识图谱回答提问。

核心能力

  • 用内置爬虫抓取官网与文档站，支持 sitemap 批量导入
  • 提供 Perplexity 风格对话式搜索页面
  • 可嵌入 JavaScript 片段，在网站右下角挂问答窗口
  • 基于 LlamaIndex 与 DSPy 构建 Graph RAG 检索链路
  • 用 TiDB 存储向量、聊天历史与 JSON 数据

容易踩的坑

项目处于早期开发阶段，接口和部署方式可能变动；需自备 TiDB（Serverless Vector Storage）作为向量库；对话与嵌入依赖外部大模型，需另行配置模型服务。

内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README

最近发版

5 个版本0.4.0
全部 release
  1. 0.4.0最新正式版2025-01-03 · 2025-01-03 09:33 UTC

    Highlights

    • Support Evaluation (beta) tool to evaluate the performance and reliability of the Chat Engine’s outputs (by @Icemap @634750802 ).
      • Current support key metrics:
        • Factual Correctness
        • Semantic Similarity
    • Support new LLM providers
    • Support new embedding model providers

    Improvements

    • Limit the upload file size via max_upload_file_size (10MB by default) parameter on site setting (by @sszgwdk @634750802 )
    • Support download the reference file in the chat page (by @Rutheniumlmw @634750802 )

    New Contributors

    If you are deploying Autoflow using docker, please follow the Upgrade guide to upgrade your Autoflow.

    Full Changelog: https://github.com/pingcap/autoflow/compare/v0.3.0...0.4.0

  2. 0.4.0rc1pre-release2024-12-31 · 2024-12-31 07:24 UTC
    查看更新说明

    What's Changed

  3. v0.3.00.3.02024-12-09 · 2024-12-09 05:16 UTC
    查看更新说明

    Release note

    • Rename project to autoflow
    • Multiple Knowledge Bases support
    • Support new embedding models providers

    For details, please check Release Notes for v0.3.0

    ‼️ Migrate from v0.2.x to v0.3.0

    If you want to upgrade autoflow from v0.2.x to v0.3.0, please check the breaking changes and migration guide before migration.

    New Contributors

    Full Changelog: https://github.com/pingcap/autoflow/compare/0.2.9...v0.3.0

  4. 0.2.02024-08-22 · 2024-08-22 00:26 UTC
    查看更新说明
  5. 0.1.02024-08-12 · 2024-08-12 08:45 UTC
    查看更新说明

    See release note and deployment guide

用户点评

登录用户的真实使用记录，可按版本和硬件筛选
共 0 条

    还没有人点评。第一个说出真实体验的人最有价值。

    README 原文

    未翻译
    在 GitHub 打开

    Backend Docker Image Version Frontend Docker Image Version E2E Status

    [!WARNING] Autoflow is still in the early stages of development. And we are actively working on it, the next move is to make it to a python package and make it a RAG solution e.g. pip install autoflow-ai. If you have any questions or suggestions, please feel free to contact us on Discussion.

    Introduction

    AutoFlow is an open source graph rag (graphrag: knowledge graph rag) based knowledge base tool built on top of TiDB Vector and LlamaIndex and DSPy.

    Features

    1. Perplexity-style Conversational Search page: Our platform features an advanced built-in website crawler, designed to elevate your browsing experience. This crawler effortlessly navigates official and documentation sites, ensuring comprehensive coverage and streamlined search processes through sitemap URL scraping.

    Image

    1. Embeddable JavaScript Snippet: Integrate our conversational search window effortlessly into your website by copying and embedding a simple JavaScript code snippet. This widget, typically placed at the bottom right corner of your site, facilitates instant responses to product-related queries.

    Image

    Deploy

    Tech Stack

    • TiDB – Database to store chat history, vector, json, and analytic
    • LlamaIndex - RAG framework
    • DSPy - The framework for programming—not prompting—foundation models
    • Next.js – Framework
    • Tailwind CSS – CSS framework
    • shadcn/ui - Design

    Contributing

    We welcome contributions from the community. If you are interested in contributing to the project, please read the Contributing Guidelines.

    License

    AutoFlow is open-source under the Apache License, Version 2.0. You can find it here.

    Contact

    You can reach out to us on Discord.

    相关项目

    同分类里 star 最高的项目
    对比全部 →
    ragflow
    infiniflow·Go

    面向智能体的 RAG 检索引擎，把文档解析、知识编译和上下文管理整合成一套基础设施，帮 Agent 稳定拿到高质量检索结果。

    检索增强生成+4检索增强引擎
    93947+24011kv1.0.02026-11-10 11:07 UTC
    LightRAG
    HKUDS·Python

    轻量级检索增强生成引擎，用图结构组织知识并支持增量更新，让 RAG 检索更快更省资源。

    检索增强生成+4检索增强引擎
    40639+495.8kv1.5.72026-11-09 16:17 UTC
    graphrag
    microsoft·Python

    基于知识图谱的模块化检索增强生成系统，把文档抽成实体关系图，让大模型在回答复杂问题时能跨文档做全局推理。

    知识图谱+4检索增强引擎
    36620+303.9kv3.3.02026-11-09 07:23 UTC
    AgentGuide
    adongwanai·MDX

    一份面向 AI Agent 开发与大模型求职的系统指南，覆盖上下文工程、高级 RAG、MCP、Agentic RL 和 LangGraph 实战，并附 1.2 万+ 面试题。

    AI Agent+4检索增强引擎
    11835+1111.1k2026-11-05 09:57 UTC