MCP Server

Rocketship includes an MCP (Model Context Protocol) server that enables AI coding agents to write better Rocketship tests by providing examples, patterns, and guidance.

Philosophy

The Rocketship MCP server is designed as a knowledgeable assistant that helps coding agents understand Rocketship testing patterns and best practices. Unlike traditional code generators, this MCP server:

Provides guidance, not files: Shows examples and patterns for agents to adapt
Maintains agent control: The coding agent creates all files and makes all decisions
Emphasizes learning: Helps agents understand Rocketship concepts deeply
Focuses on quality: Promotes E2E customer journey testing and best practices

Installation

The MCP server is published as an npm package and can be used with zero installation:

{
  "mcpServers": {
    "rocketship": {
      "command": "npx",
      "args": ["-y", "@rocketshipai/mcp-server@latest"]
    }
  }
}

Available Tools

The MCP server provides six assistant tools that provide guidance rather than generating files:

1. get_rocketship_examples

Get real examples and best practices for specific Rocketship features from the current codebase.

Available Plugins:

http - API endpoint testing with authentication and validation
delay - Timing control and wait operations
script - Custom JavaScript logic execution
sql - Database operations and validation
log - Structured logging and debugging
agent - AI-powered validation and analysis
browser - UI automation and testing
supabase - Direct Supabase API operations

Example Request:

"Show me examples of HTTP testing with step chaining"

What You Get:

Real YAML code examples from the codebase
Best practices for the specific plugin
Implementation guidance and patterns
Variable usage examples

2. suggest_test_structure

Suggests proper file structure and test organization based on current project configuration.

Project Types:

frontend - Browser-based testing with user journeys
backend - API endpoint testing
fullstack - Combined frontend and backend testing
api - Pure API testing focus
mobile - Mobile application testing

Example Request:

"I need a test structure for an e-commerce frontend project"

What You Get:

Recommended directory structure
File organization patterns
Plugin recommendations for your project type
Test flow suggestions

3. get_schema_info

Provides current schema information for validation and proper syntax.

Schema Sections:

plugins - Available plugins and their configurations
assertions - Validation patterns and types
save - Data extraction and variable storage
structure - Overall YAML test structure
full - Complete schema documentation

Example Request:

"Show me the schema for assertions and save operations"

What You Get:

Current schema validation rules
Required and optional fields
Examples of proper syntax
Compatibility information

4. get_cli_guidance

Provides current CLI usage patterns and commands from introspection.

Command Types:

run - Execute tests with various options
validate - Syntax and schema validation
structure - File structure and organization guidance

Example Request:

"How do I run tests with custom variables?"

What You Get:

Current CLI command examples
Flag explanations and usage
Common usage patterns
Version-specific information

5. get_rocketship_cli_installation_instructions

Get step-by-step instructions for installing the Rocketship CLI on different platforms.

Platform Support:

auto - Auto-detect platform (default)
macos-arm64 - macOS with Apple Silicon
macos-intel - macOS with Intel processors
linux - Linux distributions
windows - Windows systems

Example Request:

"How do I install Rocketship on macOS?"

What You Get:

Platform-specific installation commands
Available vs NOT available installation methods
Post-installation verification steps
Troubleshooting guidance
Prerequisites and dependencies

6. analyze_codebase_for_testing

Analyzes a codebase to suggest meaningful test scenarios based on available plugins.

Focus Areas:

user_journeys - End-to-end customer workflows
api_endpoints - API testing strategies
critical_paths - Business-critical functionality
integration_points - Service integration testing

Example Request:

"Analyze my React e-commerce app for testing opportunities"

What You Get:

Suggested test scenarios for your codebase
Plugin recommendations based on project type
Critical flow identification
Testing strategy recommendations

Integration Examples

With Claude Code

Add to your .mcp.json file in your project root:

{
  "mcpServers": {
    "rocketship": {
      "command": "npx",
      "args": ["-y", "@rocketshipai/mcp-server@latest"]
    }
  }
}

Then ask Claude for help:

"I need to install Rocketship and create API tests for my Express.js authentication endpoints."

Claude will use the MCP server to provide installation instructions and relevant testing examples to help you get started.

With Cursor

Go to Cursor Settings > Features > Enable Model Context Protocol
Add to your MCP configuration:

{
  "mcpServers": {
    "rocketship": {
      "command": "npx",
      "args": ["-y", "@rocketshipai/mcp-server@latest"]
    }
  }
}

Ask Cursor for guidance:

"How do I install Rocketship and structure a test for user login with database validation?"

With Windsurf

Add to your Windsurf MCP configuration (~/.codeium/windsurf/mcp_config.json):

{
  "mcpServers": {
    "rocketship": {
      "command": "npx",
      "args": ["-y", "@rocketshipai/mcp-server@latest"]
    }
  }
}

Then ask for assistance:

"Help me install Rocketship and understand how to use step chaining for a complete e-commerce checkout flow"

With Other MCP Clients

Any MCP-compatible client can use the Rocketship server. The server communicates via JSON-RPC over stdio, making it compatible with various AI assistants and development tools.

Best Practices

1. Ask for Specific Guidance

Instead of asking for complete test generation, ask for guidance on specific aspects:

Good:

"How do I install Rocketship on my platform?"
"Show me examples of API authentication testing"
"What assertions work best for user profile endpoints?"
"How should I structure an E2E checkout flow test?"

Less Effective:

"Generate all my tests"
"Create a complete test suite"

2. Learn the Patterns

Use the MCP server to understand Rocketship concepts:

Study the examples provided
Understand the reasoning behind best practices
Adapt patterns to your specific use case
Build your own expertise over time

3. Focus on Customer Journeys

The MCP server emphasizes E2E customer journey testing:

Think about complete user workflows
Test realistic user scenarios
Validate data consistency across steps
Include error and edge cases

4. Iterate and Improve

Use the validation tool to continuously improve:

Get initial structure guidance
Create your test implementation
Validate and get improvement suggestions
Refine based on feedback
Learn from the process

Example Workflow

Here's how a typical interaction works:

User: "I need to install Rocketship and test a user registration API 
       that creates a user, sends an email, and requires email verification"

Agent: *Uses get_rocketship_cli_installation_instructions*

MCP Server: *Returns platform-specific installation instructions with:*
- Installation commands for the user's platform
- Post-installation verification steps
- Troubleshooting guidance

Agent: "First, let's get Rocketship installed..."
       *Provides installation guidance*

User: "Great! Now I need help with the test structure"

Agent: *Uses get_rocketship_examples with feature_type="http"*

MCP Server: *Returns HTTP testing examples including:*
- Multi-step workflow patterns
- Email verification testing approaches
- Data validation between steps
- Best practices for user onboarding flows

Agent: "Based on these examples, let me help you create a test..."
       *Creates test file incorporating the patterns*

User: "Now I want to add database validation to ensure the user was created correctly"

Agent: *Uses get_plugin_config with plugin="sql"*

MCP Server: *Returns SQL plugin configuration examples*

Agent: *Helps add SQL validation step to the existing test*

User: "Can you review my test and suggest improvements?"

Agent: *Uses validate_and_suggest with the YAML content*

MCP Server: *Returns specific suggestions for improvement*

Agent: *Helps implement the suggested improvements*

Environment Variables

The MCP server respects these environment variables:

ROCKETSHIP_LOG: Set log level (DEBUG, INFO, ERROR)
NODE_ENV: Development/production mode

Troubleshooting

MCP Server Not Found

If your AI client can't find the MCP server:

Ensure Node.js 18+ is installed
Check your MCP configuration syntax
Restart your AI client after configuration changes
Verify the npm package is accessible

No Guidance Provided

If the MCP server isn't providing helpful guidance:

Be more specific in your requests
Provide context about what you're trying to test
Ask for specific features or patterns
Try different tool combinations

Examples Don't Match Your Use Case

If the provided examples don't fit your scenario:

Ask for multiple feature examples to combine
Request specific plugin configurations
Use the validation tool to refine your approach
Adapt the patterns to your specific needs

Security Considerations

The MCP server:

Never generates files: Only provides guidance and examples
No code access: Works with agent's existing knowledge
Read-only operation: Cannot modify your project
Local execution: All operations run in your environment
No data storage: Doesn't store or transmit your code

Future Enhancements

Planned improvements include:

Enhanced pattern library: More examples for complex scenarios
Interactive tutorials: Step-by-step guidance for common workflows
Context-aware suggestions: Better understanding of project structure
Performance optimization: Faster response times and better caching