Pytest MCP Server
Enables AI assistants to run and analyze pytest tests for desktop applications through interactive commands. Supports test execution, filtering, result analysis, and debugging for comprehensive test automation workflows.
README
Pytest MCP Server
A Model Context Protocol (MCP) server that enables AI assistants like Claude to run and analyze pytest tests for desktop applications.
Overview
This project demonstrates how to integrate pytest with AI assistants through MCP. It includes:
- Real Desktop Application: A fully functional calculator app built with tkinter
- Comprehensive Test Suite: pytest tests covering UI, functionality, and edge cases
- MCP Server: Connects pytest to AI assistants for interactive testing
Project Structure
pytest-mcp-server/
├── src/
│ ├── __init__.py
│ └── calculator_app.py # Desktop calculator application
├── tests/
│ ├── __init__.py
│ └── test_calculator.py # Pytest test suite
├── server.py # MCP server implementation
├── pyproject.toml # Project configuration
└── README.md
Installation
Prerequisites
- Python 3.10 or higher
- pip package manager
Setup
- Clone or navigate to the project directory:
cd /Users/303748/Documents/ws/pytest-mcp-server
- Create a virtual environment:
python3 -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
- Install dependencies:
pip install -e .
Or install with development dependencies:
pip install -e ".[dev]"
Running the Calculator App
To run the desktop calculator application:
python -m src.calculator_app
Or use the installed script:
calculator-app
Running Tests Manually
Run all tests:
pytest tests/
Run with verbose output:
pytest tests/ -v
Run only UI tests:
pytest tests/ -m ui
Run tests excluding slow ones:
pytest tests/ -m "not slow"
Run specific test class:
pytest tests/test_calculator.py::TestBasicOperations
Using the MCP Server
Configuration
-
Find your Claude Desktop config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
- macOS:
-
Add the MCP server configuration:
IMPORTANT NOTICE: You need to give execution rights to the run_server.sh file
{
"mcpServers": {
"pytest": {
"command": "/Users/303748/Documents/ws/pytest-mcp-server/run_server.sh",
"args": []
}
}
}
- Restart Claude Desktop
Using with Claude
Once configured, you can interact with your tests through Claude:
Example conversations:
You: "List all available tests in the calculator project"
You: "Run the UI tests for the calculator"
You: "Run only the error handling tests"
You: "Show me the results of the last test run"
You: "The addition test is failing, can you help me debug it?"
MCP Server Tools
The server provides three main tools:
1. run_tests
Executes pytest tests with specified options.
Parameters:
test_path(required): Path to test file or directorymarkers(optional): Pytest markers to filter testsverbose(optional): Show detailed output (default: true)capture(optional): Output capture method (default: "no")
2. list_tests
Lists all available tests without running them.
Parameters:
test_path(optional): Path to scan for tests (default: "tests/")
3. get_test_results
Retrieves and formats the last test run results.
Parameters: None
Test Markers
The test suite uses these markers:
@pytest.mark.ui- UI-related tests@pytest.mark.error_handling- Error handling tests@pytest.mark.slow- Slow-running tests
Development
Running Tests with Coverage
pytest tests/ --cov=src --cov-report=html
View coverage report:
open htmlcov/index.html # macOS
Code Formatting
Format code with Black:
black src/ tests/ server.py
Lint code with Ruff:
ruff check src/ tests/ server.py
Calculator Features
The calculator application supports:
- Basic operations: Addition, subtraction, multiplication, division
- Advanced operations: Square root
- UI controls: Clear (C), Clear Entry (CE), Backspace (←)
- Decimal numbers: Full decimal support
- Error handling: Division by zero, negative square roots, invalid expressions
Test Coverage
The test suite includes:
Basic Operations
- Addition
- Subtraction
- Multiplication
- Division
- Division by zero handling
Advanced Operations
- Square root calculations
- Square root of negative numbers
- Decimal number operations
UI Functions
- Clear button functionality
- Clear entry functionality
- Backspace operation
- Chaining multiple operations
Edge Cases
- Very large numbers
- Operations with zero
- Invalid expressions
Troubleshooting
Tests not running
Make sure you're in the project directory and have activated the virtual environment:
cd /Users/303748/Documents/ws/pytest-mcp-server
source venv/bin/activate
pytest tests/
MCP server not connecting
- Check that the path in
claude_desktop_config.jsonis correct - Ensure Python is in your PATH
- Restart Claude Desktop after configuration changes
- Check Claude Desktop logs for error messages
Import errors
Make sure the package is installed in editable mode:
pip install -e .
Contributing
Contributions are welcome! Areas for improvement:
- Add more calculator features (memory, history)
- Implement screenshot capture on test failures
- Add performance benchmarking tools
- Create more test scenarios
- Improve error messages
Resources
Author
Created as an example project for the LinkedIn article "Building an MCP Server for Pytest: Automating Desktop Application Testing"
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。