MCP 서버가 개발 작업을 가속화하는 데 도움이되는 워크샵

1 부: MCP에 소개

##1. Python 환경 설정

python 3.12 + 설치가 완료되었습니다.

2. 서버로 재생

가상 env 활성화 실행으로 …

터미널은 starting server] 라고 합니다.

이제 JSON-RPC 요청 줄을 줄에 붙여 넣기, 줄을 무시 #

1. 클라이언트를 초기화

2. 알림 초기화

참고: 여기에 JSON 응답을받지 않습니다.

3. 도구 목록

4. 도구가 날씨를 얻을 호출

##3. MCP 검사기에서 이러한 기능을 시각화하자

**1. mcp-inspector 설치 **

노드가 설치되지 않은 경우, brew 설치를 사용

브라우저에서 열린 MCP 검사기 http://localhost:6274/?MCP_PROXY_AUTH_TOKEN= 추가

**2. 두 가지를 참고 **

• python 환경
• 유형 which python이 같은 대답을 얻을 것이다 /mcp-for-ai-dev-course/code/.venv/bin/python
• [T5]에 경로
• vscode에서 stdio_server.py파일을 클릭하고 경로 복사 /mcp-for-ai-dev-course/code/0-mcp-demo/stdio/stdio_client.py]

YOUR_BASE_PATH]를 노트북에 따라 오른쪽 경로로 교체

** 3.

MCP Servers and AI Dev Workflows

A workshop on how MCP servers help you accelerate your development workflows

Part 1: Intro to MCP

1. Setting up Python Environment

Ensure you have python 3.12+ installed

cd code
uv sync
uv venv
source .venv/bin/activate

2. Play with your server

With your virtual env activated run…

cd 0-mcp-demo/stdio/
python stdio_server.py

The terminal will say starting server

Now paste your JSON-RPC requests line by line, ignore lines starts with #

1. Initialize the client

{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test-client","version":"1.0.0"}}}

2. initialize the notification

{"jsonrpc":"2.0","method":"notifications/initialized"}

Note: You won’t receive a JSON response here

3. list the tools

{"jsonrpc":"2.0","id":2,"method":"tools/list"}

4. call the tool get weather

{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"get_weather","arguments":{"city":"London"}}}

3. Let’s visualize these features in MCP inspector

1. Install mcp-inspector

npx @modelcontextprotocol/inspector

If you do not have node installed, use brew installation

brew install mcp-inspector
mcp-inspector

On the MCP inspector that opens on your browser http://localhost:6274/?MCP_PROXY_AUTH_TOKEN= add

2. Note down two things

• Your python environment
  • Type which pythonyou will get an answer like this /mcp-for-ai-dev-course/code/.venv/bin/python
• Path to the stdio_server.py
  • On vscode, Right click on the stdio_server.py file and copy path /mcp-for-ai-dev-course/code/0-mcp-demo/stdio/stdio_client.py

Replace YOUR_BASE_PATH with right path according to your laptop

3. Add these to MCP inspector

1. Under command - add python path
2. Under Arguments - add stdio_server path

Note: You can also run

npx @modelcontextprotocol/inspector \
  /mcp-for-ai-dev-course/code/.venv/bin/python \
  /mcp-for-ai-dev-course/code/0-mcp-demo/stdio/stdio_server.py

Part 2: MCP on AI Dev workflow

1. Create an account in context7 and copy your API Key, keep it aside

We are not affiliated with context7, just a tool that works well

2. You need Vscode + github copilot (free version should do) for this demo

3. Two ways to install MCP server mcp.json or MCP extensions

Demo workflow

1. Create an airflow pipeline without mcp servers
2. Highlight the disadvantages
3. Enable MCP servers
4. Add better instructions