KCI MCP Server

KCI (Korea Citation Index) Open API-based academic paper search and analysis MCP server

Integrates domestic academic paper search, detailed retrieval, reference analysis, and citation index queries with Claude AI using the National Research Foundation of Korea's KCI Open API.

✨ Key Features

• 7 Tools: Full support for paper search, details, references, citation indices, author search, and keyword search

• Official API Compliance: Uses open.kci.go.kr official endpoints and MetaData XML structure

• 5 KCI APIs: articleSearch / articleDetail / referenceSearch / citation / citationDetail

• STDIO + HTTP: Supports both local (Claude Desktop) and remote deployment

• TypeScript: Full type safety and structured responses

🚀 Installation and Usage

Method 1: Use directly on Claude.ai web (No installation, easiest)

No installation required; just enter one URL.

⚠️ Requires Claude Pro / Max / Team / Enterprise subscription (Free tier only supports 1 connector)

How to add a connector:

1. Log in to claude.ai.

2. Click on your name at the bottom of the left sidebar.

3. Select "Settings".

4. Go to the "Connectors" menu.

5. Click the "Add custom connector" button.

6. Enter the following details:

   • Name: kci-mcp (Any name you prefer)

   • URL: Replace YOUR_KCI_API_KEY in the address below with your KCI API key:

https://kci-mcp.fly.dev/mcp?key=YOUR_KCI_API_KEY

💡 Get KCI API Key: KCI Portal → Login → Apply for Open API Example format: https://kci-mcp.fly.dev/mcp?key=<your_KCI_API_key>

7. Click Save and return to the Claude chat window to start using it immediately.

Usage Example:

AI 교육 관련 KCI 논문 최근 10건 검색해줘
2024년 KCI 인용지수 상위 20개 학술지 알려줘

Method 2: Local installation on Claude Desktop

Prerequisites

• Node.js 18+

Installation

git clone https://github.com/auraweon7-ship-it/kci-mcp.git
cd kci-mcp
npm install

Claude Desktop Configuration

Open the configuration file at the following path:

💡 Shortcut: Claude Desktop → Settings → Developer → Click Edit Config

Add the following to your configuration file (replace with your KCI API key):

{
  "mcpServers": {
    "kci-mcp": {
      "command": "node",
      "args": ["/절대경로/kci-mcp/dist/index.js"],
      "env": {
        "KCI_API_KEY": "<발급받은_KCI_API_키>"
      }
    }
  }
}

⚠️ For Windows paths, you must use double backslashes: C:\Users\Username\kci-mcp\dist\index.js

💡 Get KCI API Key: KCI Portal → Login → Apply for Open API (Free)

Restart Claude Desktop completely, and a 🔨 icon will appear at the bottom of the chat window.

Method 3: Claude Code (Terminal, for developers)

# 전역 등록 (모든 프로젝트에서 사용, API 키를 본인 키로 교체)
claude mcp add-json kci-mcp \
  '{"type":"stdio","command":"node","args":["/절대경로/kci-mcp/dist/index.js"],"env":{"KCI_API_KEY":"<발급받은_KCI_API_키>"}}' \
  --scope user

Or create a .mcp.json file in the project root:

{
  "mcpServers": {
    "kci-mcp": {
      "command": "node",
      "args": ["./dist/index.js"],
      "env": {
        "KCI_API_KEY": "<발급받은_KCI_API_키>"
      }
    }
  }
}

✅ Verification

🛠️ Tool List (7)

💬 Example Prompts

# 논문 검색
AI 교육 관련 논문 최근 10건 검색해줘

# 저자 검색
홍길동 교수 논문 목록 찾아줘

# 키워드 검색
딥러닝 키워드 논문 2022년 이후로 검색해줘

# 인용지수 조회
2024년 KCI 인용지수 상위 20개 학술지 알려줘

# 논문 상세 조회 (검색 결과의 articleId 사용)
ART002358582 논문 상세 정보 보여줘

🔧 Run Manually (HTTP Server Mode)

# HTTP 모드로 실행 (원격 배포용)
KCI_API_KEY=<발급받은_KCI_API_키> node dist/index.js --http
# → http://localhost:3001/mcp
# → http://localhost:3001/health

📋 Environment Variables

📚 KCI Open API Information

🐳 Docker Deployment

docker build -t kci-mcp .
docker run -e KCI_API_KEY=<발급받은_KCI_API_키> -p 3001:3001 kci-mcp

📄 License

MIT

👤 Creator

Park Jeong Weon

🏫 Division of Chinese Language, Literature and Culture, Hankuk University of Foreign Studies

🏛️ Member, AI Special Committee, National Education Commission 📋 AI Talent Development Task Force, Ministry of Education

🌐 K-AI Education Network: https://k-ai.it.kr/ 📧 auraweon7@gmail.com 📱 APP: kletter.kr / kteacher.kr