Install
$ agentstack add mcp-wdf-wyh-agentic-rag-knowledge-base ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ● Environment & secrets Used
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Agentic RAG Knowledge Base
把文档变成可对话的知识系统 Local-first Agentic RAG for private / on-prem knowledge Q&A
English · Start Here · Quickstart · Enterprise · Architecture
一个面向本地部署 / 私有知识管理的 Agentic RAG 系统:文档入库 → 混合检索 → 带来源问答 → Agent 工具调用 → 可选联网搜索。
适合:想快速搭私有知识库的团队,以及想在 RAG 之上继续做 Agent / 多模型 / 企业能力的开发者。
为什么不是「又一个 RAG Demo」
| 常见 RAG 项目 | 本项目 | | --- | --- | | 检索 → 一次生成回答 | RAG + ReAct Agent,可规划、调工具、多步推理 | | 只有向量检索 | 向量 + BM25 混合检索 + Rerank | | 脚本级 PoC | Vue 3 完整前端:流式对话、来源、历史、文件管理 | | 绑定单一云厂商 | Ollama / DeepSeek / OpenAI / Gemini 可切换 | | 难二次开发 | FastAPI + MCP Server,可接入 Cursor / Claude Desktop | | 无评测 | 内置 RAG 回测(vector / bm25 / hybrid) | | 难上生产 | 多租户、JWT/OIDC、审计、配额、Prometheus/Grafana |
核心能力
- Agentic Workflow — ReAct 推理循环:规划 → 工具调用 → 汇总
- Hybrid Retrieval + Rerank — ChromaDB + BM25,bge-reranker-v2-m3 精排
- GraphRAG — 轻量知识图谱,实体关系与多跳查询
- Streaming Chat UI — 流式回答、思维过程、来源(页码 / chunk)
- Local Embedding — bge-small-zh-v1.5,可完全离线
- Incremental Index — 文件哈希增量重建
- MCP Server —
rag_search/graph_query接入 IDE Agent - RAG Evaluation — CLI + 前端一键回测
- Enterprise Ops — 多租户、审计、配额、Webhook、PII/ABAC、数据保留、合规导出、监控告警
- Web Search Ready — 可选 SearXNG / Tavily
产品预览
| 登录页 | 首页工作台 | | :---: | :---: | | | |
| 知识库构建 | 文件管理 | | :---: | :---: | | | |
| 带来源问答 | 智能 Agent | | :---: | :---: | | | |
| 设置 | | :---: | | |
Tech Stack
Frontend Vue 3 · Vite · Element Plus
Backend FastAPI · JWT / OIDC
Retrieval ChromaDB · BM25 · Reranker · GraphRAG
LLM Ollama · DeepSeek · OpenAI · Gemini
Ops Docker Compose · Prometheus · Grafana · MCP
架构一览
┌─────────────┐ ┌──────────────────────┐ ┌─────────────────┐
│ Vue 3 UI │────▶│ FastAPI + Agent │────▶│ LLM Providers │
│ Chat/KB/ │ SSE │ ReAct · Tools · MCP │ │ Ollama/Cloud │
│ Eval/Admin │◀────│ │◀────│ │
└─────────────┘ └──────────┬───────────┘ └─────────────────┘
│
┌──────────▼───────────┐
│ Hybrid Retriever │
│ Vector + BM25 + Graph│
│ + Incremental Index │
└──────────────────────┘
Quick Start
环境要求
- Python 3.10+
- Node.js 18+
- 可选:Ollama(本地模型)
1. 安装与配置
pip install -r requirements.txt
cd frontend && npm install && cd ..
cp .env.example .env
在 .env 中至少配置一种模型:
| 模式 | 关键配置 | | --- | --- | | 本地 | MODEL_PROVIDER=ollama | | 云端 | MODEL_PROVIDER=deepseek / openai / gemini + 对应 API Key |
2. 一键启动
Windows
powershell -ExecutionPolicy Bypass -File .\start.ps1
# 或双击 start.bat
macOS / Linux
bash start.sh
手动启动
python run_api.py # API → http://localhost:8000
cd frontend && npm run dev # UI → http://localhost:5173
Docker Compose
cp .env.example .env
docker compose up -d --build
# Web → http://localhost
3. 第一次使用
- 打开前端 → 设置 选择模型提供者
- 上传
md / pdf / docx / txt - 构建知识库
- 用「纯 RAG」或「智能模式」提问
4. 进阶
# RAG 回测
python run_backtest.py --build
# MCP(Cursor / Claude Desktop)
python mcp_server.py
项目结构
.
├── src/
│ ├── agent/ # Agent、工具、意图路由
│ ├── api/ # FastAPI 路由
│ ├── core/ # 向量库、检索、文档处理
│ ├── services/ # LLM、会话、RAG
│ ├── config/ # 配置
│ └── utils/ # 日志、监控、重试
├── frontend/ # Vue 3 前端
├── deploy/ # Docker / 监控 / SearXNG
├── docs/ # 文档与截图
├── documents/ # 默认知识源
└── vector_db/ # 向量持久化
常用配置
详见 [.env.example](.env.example):
MODEL_PROVIDER/DEEPSEEK_API_KEY/OLLAMA_MODEL/OLLAMA_API_URLVECTOR_DB_PATH/TOP_K/MAX_TOKENS- 联网搜索:
TAVILY_API_KEY
文档
| 文档 | 说明 | | --- | --- | | [Start Here](STARTHERE.md) | 最短上手路径 | | [Quickstart](QUICKSTART.md) | 快速开始 | | [Agent 架构](docs/AGENTARCHITECTURE.md) | ReAct / 工具路由 | | [企业部署](docs/ENTERPRISEDEPLOYMENT.md) | 多租户与运维 | | [Demo 素材清单](docs/DEMOASSETSCHECKLIST.md) | 截图 / GIF | | [日志排查](LOGQUICK_GUIDE.md) | 排障 |
English
Agentic RAG Knowledge Base is a local-first knowledge Q&A system for private deployments:
- Hybrid retrieval (vector + BM25) with reranking
- ReAct agent with tool calling and optional web search
- Vue 3 streaming chat UI with citations and history
- Multi-provider LLMs: Ollama, DeepSeek, OpenAI, Gemini
- MCP server for Cursor / Claude Desktop
- Built-in RAG evaluation and enterprise ops hooks
pip install -r requirements.txt
cd frontend && npm install && cd ..
cp .env.example .env
# set MODEL_PROVIDER + API key (or Ollama)
bash start.sh # or start.ps1 / start.bat on Windows
Open http://localhost:5173, upload documents, build the index, and chat.
Roadmap
- 更稳的默认配置与新手引导
- HuggingFace Spaces / 在线 Demo
- 更完整的英文文档与示例知识库
- Agent 工具生态与插件市场
欢迎 Star / Issue / PR。如果你在私有化或中文知识库场景落地了,也欢迎分享反馈。
License
[MIT](LICENSE)
Source & license
This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: wdf-wyh
- Source: wdf-wyh/Agentic-RAG-Knowledge-Base
- License: MIT
- Homepage: https://github.com/wdf-wyh/Agentic-RAG-Knowledge-Base
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.