VisitKorea MCP Server

An open-source Model Context Protocol (MCP) server that connects AI agents — including Claude Desktop and Manus AI — to the Korea Tourism Organization (KTO) English Open Data API. Ask any MCP-compatible AI about Korean tourist attractions, restaurants, accommodations, festivals, and more, in English.

API Data Source: Korea Tourism Organization English Open API

This is also where you apply for and obtain your API key.

Architecture

AI Agent (Claude / Manus AI)
        │  MCP over HTTP (Streamable HTTP)
        ▼
Node.js Express Proxy  (port 8080, path /mcp)
        │  localhost:3001
        ▼
Python MCP Server  (Starlette + uvicorn, port 3001)
        │  HTTPS
        ▼
Korea Tourism Organization Open API
(apis.data.go.kr/B551011/EngService2)

The Node.js server acts as the public-facing entry point (handling Replit's routing and deployment lifecycle). It spawns the Python MCP server as a subprocess and proxies all /mcp requests to it.

Features

14 MCP Tools

| Tool | Description | |------|-------------| | search_tourism_by_area | Search attractions, restaurants, accommodations, shopping, cultural facilities, leisure spots, and festivals by province or district | | search_tourism_by_location | Find tourism spots near a GPS coordinate within a specified radius — ideal for "things to do near me" queries | | search_tourism_by_keyword | Full-text keyword search across all content types in English | | search_festivals_and_events | Discover Korean festivals, performances, and events by date range | | search_accommodations | Browse hotels, pensions, hostels, condominiums, and camping sites | | get_tourism_common_info | Retrieve title, address, GPS, phone, homepage, and overview for any content item by ID | | get_tourism_intro_info | Get type-specific details: heritage info for attractions, check-in/out for hotels, menu specialties for restaurants, etc. | | get_tourism_detail_info | Get repeating sub-data: room types with pricing for hotels, menu items for restaurants, program schedules for events | | get_tourism_images | Fetch all photo URLs (original + thumbnail) associated with a content item | | get_sync_list | List content items created or modified since a given date — useful for keeping local databases in sync | | get_legal_district_codes | Look up province and city/county codes (법정동) used to filter search results | | get_classification_codes | Browse the new 3-level classification hierarchy (AC=Accommodation, EV=Events, FO=Food, VE=Culture, etc.) | | get_area_codes | Look up the legacy area codes (17 provinces) used by older API parameters | | get_category_codes | Browse the legacy 3-level category hierarchy (A01=Nature, A02=Culture, A05=Cuisine, etc.) |

Content Types

| ID | Type | |----|------| | 75 | Leisure / Sports | | 76 | Tourist Attraction | | 78 | Cultural Facility | | 79 | Shopping | | 80 | Accommodation | | 82 | Restaurant / Food | | 85 | Festival / Performance / Event |

Prerequisites

• A Korea Tourism Organization API key from data.go.kr
  • Register for a free account, apply for the "영문 관광정보 서비스" (English Tourism Information Service), and copy the URL-encoded service key from your application.
• A Replit account (free tier works)

Local Setup

1. Clone the repository

git clone https://github.com/leejaew/custom-visitkorea-mcp.git
cd custom-visitkorea-mcp

2. Install Node.js dependencies

npm install -g pnpm
pnpm install

3. Install Python dependencies

pip install -r visitkorea-mcp/requirements.txt

4. Set the API key environment variable

export VISITKOREA_API_KEY="your_url_encoded_service_key_here"

The service key from data.go.kr is already URL-encoded. Copy it exactly as shown on your application page.

5. Run the Python MCP server directly (stdio mode — for Claude Desktop)

python3 visitkorea-mcp/server.py

6. Run the Python MCP server in HTTP mode (for web-based clients)

python3 visitkorea-mcp/server.py --http --port 3001

The server will be available at http://localhost:3001/mcp.

Connecting to Claude Desktop

Add the following to your Claude Desktop claude_desktop_config.json:

{
  "mcpServers": {
    "visitkorea": {
      "command": "python3",
      "args": ["/absolute/path/to/visitkorea-mcp/server.py"],
      "env": {
        "VISITKOREA_API_KEY": "your_url_encoded_service_key_here"
      }
    }
  }
}

Connecting to Manus AI

After deploying your own instance to Replit (see the deployment steps below), use your own endpoint:

https://<your-repl-name>.replit.app/mcp

In Manus AI, add a new MCP server with this URL. No authentication token is required — the API key is handled server-side.

Deploying to Replit

Follow these steps to deploy your own fully functional instance of this MCP server on Replit.

New to Replit? Sign up for a free account using this referral link: https://replit.com/refer/leejaew

Step 1 — Import the repository

1. Go to replit.com and sign in.
2. Click Create Repl → Import from GitHub.
3. Paste the repository URL: https://github.com/leejaew/custom-visitkorea-mcp
4. Replit will detect the workspace configuration automatically.

Step 2 — Add your API key as a Secret

1. In your Repl, open the Secrets panel (lock icon in the sidebar, or Tools → Secrets).
2. Create a new secret:
   • Key: VISITKOREA_API_KEY
   • Value: Your URL-encoded service key from data.go.kr

Do not paste the key into any source file. Always use the Secrets panel so it is never committed to version control.

Step 3 — Run the development server

Click the Run button (or press the green Play button). Replit will:

1. Install all Node.js and Python dependencies automatically.
2. Build the Node.js proxy server.
3. Start both the Node.js server (port 8080) and the Python MCP server (port 3001).

Step 4 — Verify it works

Open the Replit Shell and run:

curl -X POST http://localhost:8080/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

You should receive a JSON response listing all 14 tools.

Step 5 — Deploy to production

1. Click Deploy in the top-right corner of Replit.
2. Choose Autoscale deployment.
3. Click Deploy to publish.

Your MCP server will be publicly accessible at:

https://<your-repl-name>.replit.app/mcp

Environment Variables

| Variable | Required | Description | |----------|----------|-------------| | VISITKOREA_API_KEY | Yes | URL-encoded service key from data.go.kr | | PORT | Set by Replit | Port for the Node.js proxy (default 8080 in dev, assigned by Replit in production) |

Project Structure

├── visitkorea-mcp/
│   ├── server.py          # Python MCP server (all 14 tools)
│   └── requirements.txt   # Python dependencies
├── artifacts/
│   └── api-server/
│       ├── src/
│       │   ├── app.ts     # Express app with wait middleware + MCP proxy
│       │   ├── index.ts   # Entry point — spawns Python server, starts Express
│       │   ├── lib/
│       │   │   └── logger.ts
│       │   └── routes/
│       │       ├── index.ts
│       │       └── health.ts
│       ├── build.mjs      # esbuild bundler config
│       ├── package.json
│       └── tsconfig.json
├── lib/                   # Shared TypeScript libraries
│   ├── api-spec/          # OpenAPI specification
│   ├── api-zod/           # Generated Zod schemas
│   ├── db/                # Drizzle ORM database layer
│   └── api-client-react/  # Generated React hooks
├── .replit                # Replit workspace configuration
├── pnpm-workspace.yaml    # pnpm monorepo configuration
└── README.md

Data Source

This project uses the Korea Tourism Organization English Open Data API:

• Portal: https://www.data.go.kr/data/15101753/openapi.do
• Provider: Korea Tourism Organization (한국관광공사) via the Korean Public Data Portal (공공데이터포털)
• API Base URL: https://apis.data.go.kr/B551011/EngService2
• Language: English
• License: Public Data, free for commercial and non-commercial use with attribution

To obtain an API key:

1. Visit data.go.kr
2. Sign up for a free account
3. Click 활용신청 (Apply for Use) on the dataset page
4. After approval (usually instant), copy your 인증키 (Service Key) — use the URL-encoded version

License

MIT License — open source, free to use, modify, and distribute.

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.