mcp-server-demo
MCP 서버 데모 (Python)
두 가지 도구를 갖춘 처음부터 만드는 로컬 MCP 서버:
weather(city)→ wttr.in을 통한 현재 날씨query_db(sql)→ 읽기 전용 SQLite SELECT 쿼리
1) 설정
python -m venv .venv
source .venv/bin/activate
pip install -U pip
pip install -e .Related MCP server: Swiss Meteo MCP
2) MCP 서버 실행
mcp-server-demo이 명령은 로컬 서버를 시작하고 샘플 행이 포함된 demo.db를 자동으로 생성합니다.
3) 빠른 수동 MCP 확인 (선택 사항)
MCP Inspector가 설치되어 있는 경우:
npx @modelcontextprotocol/inspector그런 다음 로컬 Python 서버에 연결합니다.
4) OpenAI API 통합 옵션
env 파일을 복사하고 키를 입력합니다:
cp .env.example .env한 터미널에서 서버를 시작합니다.
두 번째 터미널에서 다음을 실행합니다:
python client_openai_api.py참고: 이 작업은
http://localhost:8000/mcp를 가리키는 MCP 도구 정의와 함께 OpenAI Responses API를 사용합니다.
도구 동작
weather(city: str)
온도, 체감 온도, 습도, 바람, 간단한 상태를 포함한 JSON 요약 필드를 반환합니다.
query_db(sql: str)
SELECT ...쿼리만 허용합니다.행을 JSON으로 반환합니다.
이 시작 데모에서는 안전을 위해 SELECT가 아닌 SQL은 거부합니다.
프로젝트 파일
server.py— FastMCP 서버 + 도구 정의.client_openai_api.py— MCP 도구를 호출할 수 있는 간단한 OpenAI API 호출.pyproject.toml— 종속성 + 스크립트 진입점.
Available Tools
10 toolsanalyze_image_with_openaiC
Analyze an image file with an OpenAI vision-capable model.
Inputs: path (required), prompt (required), model. Returns: result
inspect_fileC
Return file metadata and content preview for text/csv/image workflows.
Inputs: path (required), preview_chars, include_base64. Returns: result
list_directoriesC
List directories directly inside a folder within MCP_FILE_OPS_ROOT.
Inputs: path. Returns: result
list_filesB
List files directly inside a folder within MCP_FILE_OPS_ROOT.
Inputs: path. Returns: result
make_directoryC
Create a directory inside MCP_FILE_OPS_ROOT.
Inputs: path (required). Returns: result
move_fileC
Move a file from source_path to destination_path inside MCP_FILE_OPS_ROOT.
Inputs: source_path (required), destination_path (required). Returns: result
move_files_by_globC
Move all files matching a glob pattern from source_dir into destination_dir.
Inputs: source_dir (required), pattern (required), destination_dir (required). Returns: result
query_dbA
Run a read-only SELECT query against local SQLite demo.db.
Inputs: sql (required). Returns: result
read_fileC
Read a UTF-8 text file inside MCP_FILE_OPS_ROOT.
Inputs: path (required). Returns: result
weatherB
Return current weather for a city using wttr.in.
Inputs: city (required). Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
Most tools have distinct purposes, but there is some overlap between inspect_file and read_file (both provide file content), and move_file vs move_files_by_glob are similar in name and action.
Majority follow verb_noun pattern with underscores, but 'weather' is a lone noun and 'analyze_image_with_openai' is verbose and breaks the pattern.
10 tools is a reasonable number, but the server covers multiple unrelated domains (files, DB, weather, image), making it feel a bit scattered.
File operations lack delete/copy, DB only has read-only query, weather and image analysis are single-tool services. The surface is incomplete for any single domain.
Related MCP Connectors
An MCP server for weather information by @kulybaba
An MCP server for weather information by @kulybaba
Related MCP Servers
- FlicenseBqualityDmaintenanceA demonstration server showcasing MCP capabilities with basic tools including addition calculations and weather API integration for fetching city weather data.2-
- FlicenseCqualityDmaintenanceAn experimental MCP server that provides weather information tools, including temperature lookup via wttr.io, along with note-taking capabilities for learning MCP server development with GitHub Copilot.3-
- FlicenseNot gradedqualityDmaintenanceA demo MCP server that provides AI assistants access to live US weather data via the free National Weather Service API, offering forecast and alert tools without requiring an API key.-
- FlicenseNot gradedqualityDmaintenanceA lightweight MCP server providing tools for adding integers, getting current time, and fetching weather forecasts via wttr.in.-