快速接入

先看主来源和安装命令，再打开仓库或文档；下面只保留这个条目的关键接入事实。

详细介绍

DB Knowledge MCP

Enterprise-grade MCP server untuk membaca metadata database dan menyimpan ke vector knowledge base dengan UI graph interaktif (Obsidian-like).

🎯 Overview

DB Knowledge MCP membaca metadata database (SQL Server, MySQL, PostgreSQL) dan menyimpannya sebagai vector knowledge base yang bisa dibaca cepat oleh AI agent. Dilengkapi dengan UI graph interaktif seperti Obsidian untuk visualisasi dan eksplorasi relationship antar tabel.

✨ Features

Core Features

Multi-Database Support: SQL Server, MySQL, PostgreSQL
Vector Knowledge Base: Simpan schema sebagai embedded vectors
8 MCP Tools: list_databases, scan_database, search_knowledge, dll
Semantic Search: Query natural language ke knowledge base

UI Dashboard (Port 5005)

Interactive Graph: Vis Network visualization
Clickable Nodes: Klik node untuk melihat detail
Database Selector: Switch antar knowledge bases
Real-time Search: Filter nodes dengan search bar
Zoom & Pan: Navigasi graph dengan mouse
Drag Nodes: Atur posisi node sesuka hati
Auto-layout: Button untuk auto-arrange graph

🚀 Quick Start

Prerequisites

Node.js >= 18.0.0
Database credentials (SQL Server, MySQL, atau PostgreSQL)

Installation

git clone https://github.com/PTRJRepository/db-knowledge-mcp.git
cd db-knowledge-mcp
npm install
npm run build

Configuration

Edit config/databases.json:

{
  "databases": [
    {
      "id": "db_payroll",
      "name": "Payroll Database",
      "type": "sqlserver",
      "host": "localhost",
      "port": 1433,
      "database": "payroll_db",
      "username": "sa",
      "password": "your_password",
      "enabled": true,
      "description": "Production payroll system"
    },
    {
      "id": "db_hris",
      "name": "HRIS Database",
      "type": "mysql",
      "host": "localhost",
      "port": 3306,
      "database": "hris_db",
      "username": "root",
      "password": "your_password",
      "enabled": true
    }
  ],
  "embedder": {
    "provider": "openai",
    "model": "text-embedding-3-small",
    "dimensions": 1536
  },
  "storage": {
    "type": "json",
    "path": "knowledgebases"
  }
}

Running

Mode 1: Combined Server (REST API + UI Dashboard)

npm run start:combined

UI Dashboard: http://localhost:5005
REST API: http://localhost:5005/api/*

Mode 2: MCP Server Only (STDIO)

npm start

Berguna untuk integrate dengan AI tools lain (Zo Computer, Claude, dll)

Mode 3: Scan All Databases

npm run scan-all

📊 UI Dashboard Guide

Graph View

┌─────────────────────────────────────────────────────────────┐
│  DB Knowledge Graph                              [Search...] │
├─────────────────────────────────────────────────────────────┤
│  [Dropdown: Select Database ▼]  [Scan]  [Refresh]  [Layout] │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│         ┌──────────────┐                                   │
│         │ db_payroll   │ ← Database Node (Cyan)            │
│         └──────┬───────┘                                   │
│                │                                            │
│         ┌──────┴───────┐                                   │
│         │  employees   │ ← Table Node (Green)              │
│         └──────┬───────┘                                   │
│                │                                            │
│    ┌───────────┼───────────┐                               │
│    │           │           │                               │
│ ┌──┴──┐   ┌───┴───┐   ┌───┴───┐                           │
│ │ id  │   │name   │   │ dept  │ ← Column Nodes (Yellow)   │
│ └──┬──┘   └───┬───┘   └───┬───┘                           │
│    │         │           │                                 │
│    └─────────┴───────────┘                                 │
│    │                                           (Click node)  │
│    └────────────► Shows detail panel ◄────────────         │
│                                                             │
├─────────────────────────────────────────────────────────────┤
│ Legend: ● Database  ● Table  ● Column  ● Relationship      │
└─────────────────────────────────────────────────────────────┘

Node Colors

Type	Color	Description
Database	Cyan (#00bcd4)	Root node untuk setiap database
Table	Green (#4caf50)	Tabel dalam database
Column	Yellow (#ffeb3b)	Kolom dalam tabel
Relationship	Red (#f44336)	Foreign key relationship

Using the UI

Select Database: Gunakan dropdown untuk memilih database
Scan: Klik "Scan" untuk scan database dan update knowledge base
Search: Ketik di search bar untuk filter nodes
Click Node: Klik node untuk melihat detail di panel kanan
Drag: Drag node untuk reposisi
Zoom: Scroll mouse untuk zoom in/out
Layout: Klik "Layout" untuk auto-arrange graph

Detail Panel

Ketika node diklik, panel kanan menampilkan:

Database Node:

┌─────────────────────────┐
│ 📊 db_payroll           │
├─────────────────────────┤
│ Type: SQL Server        │
│ Tables: 25              │
│ Columns: 156            │
│ Last Updated: 2026-03  │
├─────────────────────────┤
│ [Scan] [Delete]         │
└─────────────────────────┘

Table Node:

┌─────────────────────────┐
│ 📋 employees           │
├─────────────────────────┤
│ Schema: dbo             │
│ Columns: 12             │
│ Row Count: 1,250        │
├─────────────────────────┤
│ Primary Keys:           │
│   • employee_id         │
├─────────────────────────┤
│ Foreign Keys:            │
│   • department_id →     │
│     departments(id)     │
├─────────────────────────┤
│ [View Details]          │
└─────────────────────────┘

Column Node:

┌─────────────────────────┐
│ 📌 employee_id          │
├─────────────────────────┤
│ Type: int               │
│ Nullable: No             │
│ Primary Key: Yes        │
│ Foreign Key: No         │
├─────────────────────────┤
│ Part of: employees      │
└─────────────────────────┘

🔧 MCP Tools

1. list_databases

List semua database yang dikonfigurasi.

{
  "name": "list_databases",
  "description": "List all configured databases"
}

2. scan_database

Scan database dan simpan ke knowledge base.

{
  "name": "scan_database",
  "arguments": {
    "databaseId": "db_payroll"
  }
}

3. search_knowledge

Search knowledge base dengan query.

{
  "name": "search_knowledge",
  "arguments": {
    "query": "employee salary",
    "databaseId": "db_payroll",
    "options": {
      "limit": 10,
      "minScore": 0.1
    }
  }
}

4. get_table_info

Get detail table.

{
  "name": "get_table_info",
  "arguments": {
    "databaseId": "db_payroll",
    "tableName": "employees",
    "schema": "dbo"
  }
}

5. get_column_info

Get detail column.

{
  "name": "get_column_info",
  "arguments": {
    "databaseId": "db_payroll",
    "tableName": "employees",
    "columnName": "salary"
  }
}

6. get_table_relationships

Get semua foreign key relationships untuk table.

{
  "name": "get_table_relationships",
  "arguments": {
    "databaseId": "db_payroll",
    "tableName": "employees"
  }
}

7. get_database_overview

Get overview database termasuk semua tables.

{
  "name": "get_database_overview",
  "arguments": {
    "databaseId": "db_payroll"
  }
}

8. delete_knowledge_base

Delete knowledge base untuk database.

{
  "name": "delete_knowledge_base",
  "arguments": {
    "databaseId": "db_payroll"
  }
}

🌐 REST API

Base URL

http://localhost:5005/api

Endpoints

GET /api/knowledge-bases

List semua knowledge bases.

curl http://localhost:5005/api/knowledge-bases

GET /api/knowledge-bases/:id

Get detail knowledge base.

curl http://localhost:5005/api/knowledge-bases/db_payroll

GET /api/knowledge-bases/:id/graph

Get graph data untuk visualization.

curl http://localhost:5005/api/knowledge-bases/db_payroll/graph

Response:

{
  "nodes": [
    { "id": "db_payroll", "label": "Payroll DB", "type": "database", "color": "#00bcd4" },
    { "id": "employees", "label": "employees", "type": "table", "color": "#4caf50" },
    { "id": "employee_id", "label": "employee_id", "type": "column", "color": "#ffeb3b" }
  ],
  "edges": [
    { "from": "employees", "to": "employee_id", "label": "pk" }
  ]
}

GET /api/knowledge-bases/:id/search?q=query

Search dalam knowledge base.

curl "http://localhost:5005/api/knowledge-bases/db_payroll/search?q=salary"

POST /api/scan/:databaseId

Scan database.

curl -X POST http://localhost:5005/api/scan/db_payroll

POST /api/scan-all

Scan semua database.

curl -X POST http://localhost:5005/api/scan-all

GET /api/databases

List database configs.

curl http://localhost:5005/api/databases

POST /api/test-connection/:databaseId

Test koneksi database.

curl -X POST http://localhost:5005/api/test-connection/db_payroll

GET /api/health

Health check.

curl http://localhost:5005/api/health

📁 Project Structure

db-knowledge-mcp/
├── config/
│   └── databases.json       # Konfigurasi database
├── public/
│   └── index.html           # UI Dashboard
├── src/
│   ├── connectors/          # Database connectors
│   │   ├── base.ts          # Base connector class
│   │   ├── sqlserver.ts     # SQL Server connector
│   │   ├── mysql.ts         # MySQL connector
│   │   ├── postgresql.ts    # PostgreSQL connector
│   │   └── index.ts         # Factory function
│   ├── embedder/
│   │   └── schema-embedder.ts  # Schema → Embeddings
│   ├── storage/
│   │   └── vector-store.ts  # JSON vector store
│   ├── mcp-server.ts        # MCP stdio server
│   └── server.ts            # Express + UI server
├── knowledgebases/           # Generated KB storage
├── dist/                     # Compiled output
├── docs/
│   └── API.md              # Detailed API docs
├── CLAUDE.md               # Developer guide
├── README.md               # This file
└── package.json

🔌 Database Support

SQL Server

{
  "id": "db_payroll",
  "name": "Payroll DB",
  "type": "sqlserver",
  "host": "localhost",
  "port": 1433,
  "database": "payroll",
  "username": "sa",
  "password": "***"
}

MySQL

{
  "id": "db_hris",
  "name": "HRIS DB",
  "type": "mysql",
  "host": "localhost",
  "port": 3306,
  "database": "hris",
  "username": "root",
  "password": "***"
}

PostgreSQL

{
  "id": "db_analytics",
  "name": "Analytics DB",
  "type": "postgresql",
  "host": "localhost",
  "port": 5432,
  "database": "analytics",
  "username": "postgres",
  "password": "***"
}

🔐 Security Notes

No credentials stored in code: Semua credentials di config/databases.json
Read-only access: Connector hanya baca metadata, tidak write
Local storage: Knowledge base disimpan lokal, tidak dikirim ke cloud
HTTPS-ready: Bisa taruh di reverse proxy (nginx, Caddy) untuk production

🛠️ Development

Build

npm run build

Run Tests

npm test

Lint

npm run lint

📝 License

MIT License - lihat LICENSE

🤝 Contributing

Fork repository
Buat branch baru (git checkout -b feature/xxx)
Commit changes (git commit -m 'Add feature xxx')
Push ke branch (git push origin feature/xxx)
Buat Pull Request

📧 Contact

Author: attapangestu
GitHub: https://github.com/PTRJRepository/db-knowledge-mcp

目录标签

TypeScriptClaude数据分析数据库元数据本地部署知识图谱向量存储交互式可视化SQL分析

支持客户端

Claude

接入字段

传输方式（transport，传输协议）

未说明

鉴权方式（authType，认证方式）

none

工具数量（toolCount，工具数）

8

资源数量（resourceCount，资源数）

0

提示词数量（promptCount，提示词数）

0

权限和风险

未说明none部署方式未说明

接入前请确认传输方式、认证方式和部署位置，并根据实际工具能力限制访问范围。

安装前确认

不要直接授予不必要的文件、网络或账号权限；先核对安装命令和配置内容。

仍需确认：installCommand

来源信息

MCP World

mcpworld.com

/detail/38b7b92bc8fae43ae69187c722aa03ea

DB Knowledge MCP是一个企业级工具，用于读取数据库元数据并存储到向量知识库中，提供交互式图形界面进行可视化和探索表关系。

继续浏览同类 MCP

更多数据服务 MCP（6,870）返回 MCP 目录

上一个 MCP

Dbhub MCP (Cn Dvn)

DBHub MCP Server是一个基于Go语言编写的模型上下文协议服务器，提供对MySQL和PostgreSQL的安全只读数据库操作，支持STDIO和HTTP两种传输方式，确保数据访问的安全性和高效性。

下一个 MCP

db MCP

SQLite MCP Server是一个功能丰富的SQLite数据库管理工具，提供139种专业工具、17种资源和10个AI驱动的提示，支持HTTP/SSE传输、OAuth 2.1认证、工具过滤和细粒度访问控制，适用于数据库开发和管理场景。

Db Knowledge MCP