Karate MCP Server

A Model Context Protocol (MCP) server that provides AI agents with comprehensive knowledge of the Karate DSL framework for API testing, UI automation, and performance testing.

Overview

This MCP server enables AI assistants like Claude, GitHub Copilot, and Roo Code to access deep, structured knowledge about Karate DSL, including:

• HTTP Methods - URL, path, request, headers, parameters, cookies, multipart uploads
• Response Handling - Status codes, response body, headers, cookies, timing
• Assertions - Match operators and 18+ fuzzy matchers (#string, #number, #regex, etc.)
• Data Structures - Variables, JSON, XML, YAML, CSV, tables
• JavaScript API - 13+ utility functions for data manipulation
• Data-Driven Testing - Scenario Outline, dynamic Examples, CSV/JSON data
• Code Reusability - Call, callonce, JavaScript functions
• Configuration - Timeouts, SSL, proxy, headers, logging
• Parallel Execution - Multi-threaded test execution
• UI Automation - Cross-browser web testing
• Performance Testing - Gatling integration
• API Mocking - Test doubles and consumer contracts
• Best Practices - 15+ recommended patterns

Features

Available Tools

The MCP server exposes 5 tools:

1. list_karate_categories - List all 14 Karate DSL categories
2. search_karate_features - Search features within a category
3. get_karate_feature - Get detailed info about a specific feature
4. generate_karate_example - Generate complete test examples
5. get_karate_best_practices - Get best practices and recommendations

Knowledge Base

• 14 categories of Karate features
• 100+ features and functions documented
• Syntax examples for every feature
• 6 test templates (GET, POST, Auth, Data-driven, Parallel, UI)
• 15+ best practices for maintainable tests

Installation

Prerequisites

• Node.js 18 or higher
• npm or yarn

From Source

git clone https://github.com/vspaswin/karate-mcp-server.git
cd karate-mcp-server
npm install
npm run build

Usage

With Claude Desktop

Add to your Claude Desktop configuration:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "karate": {
      "command": "node",
      "args": ["/absolute/path/to/karate-mcp-server/dist/index.js"]
    }
  }
}

Restart Claude Desktop to connect.

With VS Code (GitHub Copilot)

📖 Complete VS Code Setup Guide

Quick Start

1. Open Command Palette (Cmd+Shift+P / Ctrl+Shift+P)
2. Run: MCP: Add Server
3. Select Local (stdio)
4. Name: karate-dsl
5. Command: node
6. Args: /absolute/path/to/karate-mcp-server/dist/index.js
7. Scope: Global or Workspace

Or create .vscode/mcp.json:

{
  "servers": {
    "karate-dsl": {
      "type": "stdio",
      "command": "node",
      "args": ["/absolute/path/to/karate-mcp-server/dist/index.js"]
    }
  }
}

Test: Ask GitHub Copilot: @karate-dsl List all categories

With Roo Code (VS Code Extension)

📖 Complete Roo Code Setup Guide

Quick Start

1. Install Roo Code extension in VS Code
2. Open Roo Code panel → Click MCP icon (🔌)
3. Click "Edit Global MCP"
4. Add configuration:

{
  "mcpServers": {
    "karate-dsl": {
      "command": "node",
      "args": ["/absolute/path/to/karate-mcp-server/dist/index.js"]
    }
  }
}

5. Save and verify connection in MCP panel

Test: Ask Roo: List all Karate DSL categories

Example Queries

Learning Karate

List all Karate DSL categories
What are all the assertion operators in Karate?
Show me how to use fuzzy matchers like #string and #regex
How do I configure HTTP timeouts?
Explain callonce vs call in Karate

Generating Tests

Generate a complete POST API test example
Show me how to implement authentication with callonce
Create a data-driven test with CSV data
Generate a parallel execution example
Create a UI automation test for login

Building Real Tests

Help me write a Karate test for my REST API at https://api.example.com
I need to test a POST /users endpoint that creates users
How do I assert response time is under 2 seconds?
Show me how to extract and reuse values from responses

Best Practices

What are the best practices for Karate test organization?
How should I structure my Karate project?
Show me patterns for reusing authentication logic

Tool Examples

List All Categories

Query: "List all Karate categories"

Response: Shows all 14 categories with descriptions and item counts

Search Features

Query: "Show me all assertion features"

Tool:

{"name": "search_karate_features", "arguments": {"category": "assertions"}}

Response: 18 assertion features with syntax examples

Get Specific Feature

Query: "How does match contains work?"

Tool:

{"name": "get_karate_feature", "arguments": {"featureName": "match contains"}}

Response: Detailed feature documentation

Generate Example

Query: "Generate a POST API example"

Tool:

{"name": "generate_karate_example", "arguments": {"useCase": "api_post"}}

Response: Complete, runnable test code

Available Test Templates

• api_get - GET request with assertions and fuzzy matchers
• api_post - POST request with JSON body
• api_auth - Authentication flow with callonce for token reuse
• data_driven - Data-driven test with Scenario Outline and Examples
• parallel_test - Parallel execution with JUnit runner
• ui_test - Web UI automation with driver

Development

Build

npm run build

Watch Mode

npm run dev

Project Structure

karate-mcp-server/
├── src/
│   ├── index.ts              # MCP server with 5 tools
│   └── karate-knowledge.ts   # Karate DSL knowledge base
├── docs/
│   ├── vscode-setup.md       # VS Code/Copilot setup guide
│   └── roocode-setup.md      # Roo Code setup guide
├── dist/                     # Compiled JavaScript
├── package.json
├── tsconfig.json
├── .gitignore
└── README.md

Categories Covered

1. http_methods (11 features) - URL, path, method, request, params, headers, cookies, forms, multipart, SOAP, retry
2. response_handling (6 features) - Status, response, bytes, headers, cookies, timing
3. assertions (18 features) - Match operators, fuzzy matchers (#string, #number, #regex, etc.)
4. variables (8 features) - def, text, table, YAML, CSV, JSON, XML, copy
5. actions (10 features) - assert, print, get, set, remove, configure, call, eval, read
6. javascript_api (13 functions) - log, get, set, jsonPath, map, filter, merge, etc.
7. data_driven (4 features) - Scenario Outline, dynamic Examples, CSV/JSON data
8. reusability (4 features) - call, callonce, JavaScript functions, parameters
9. configuration (5 settings) - Timeouts, SSL, proxy, headers
10. parallel_execution (3 features) - Runner, thread configuration, @parallel tag
11. ui_automation (4 features) - driver, input, click, screenshot
12. performance_testing (2 features) - Gatling integration, load profiles
13. mocking (2 features) - Mock servers, stateful mocks
14. best_practices (15 tips) - Organization, naming, performance, maintainability

Supported AI Assistants

• Claude Desktop - Native MCP support
• GitHub Copilot (VS Code) - MCP integration in VS Code 1.102+
• Roo Code (VS Code) - AI coding assistant with MCP support
• Any MCP Client - Compatible with any stdio-based MCP client

Technical Details

• Built with @modelcontextprotocol/sdk v1.0+
• Uses stdio transport for local integration
• Type-safe with Zod schema validation
• Comprehensive knowledge base with 100+ features
• 6 ready-to-use test templates
• Works with Claude, Copilot, Roo Code, and other MCP clients

Troubleshooting

Server Not Connecting

1. Verify path to dist/index.js is absolute and correct
2. Run npm run build to ensure server is compiled
3. Check Node.js version: node --version (should be 18+)
4. Restart your AI assistant / IDE

Tools Not Available

1. Verify server shows as "Connected" in MCP panel
2. Check configuration file syntax is valid JSON
3. Look for error messages in output/console logs
4. Try asking directly: "List karate categories"

Permission Issues

1. Make dist/index.js executable: chmod +x dist/index.js
2. Ensure Node.js has execute permissions
3. On Windows, use full path: C:/Program Files/nodejs/node.exe

Contributing

Contributions welcome! Areas for improvement:

• Additional test examples and templates
• More detailed feature explanations
• CI/CD integration examples
• Video tutorials and demos
• Support for more AI assistants

Resources

• Karate DSL Official Repo
• Karate Documentation
• Model Context Protocol
• MCP TypeScript SDK
• VS Code MCP Documentation
• Roo Code Documentation

License

MIT

Author

vspaswin
Lead Software Engineer | Test Automation Expert

Acknowledgments

• Karate DSL team for the outstanding test automation framework
• Anthropic for the Model Context Protocol specification
• Microsoft & GitHub for VS Code and Copilot MCP integration
• Roo Code team for excellent AI coding assistant
• MCP community for tools and examples

---

Note: This is an independent community project, not officially affiliated with Karate Labs, Intuit, Microsoft, GitHub, or Anthropic.

Test update