Use Existing MCP Servers
Learn how to integrate popular MCP servers like Claude, Cursor, and Windsurf into your applications.
Using pre-built MCP servers
In this guide, you'll learn how to:
- Connect to public MCP servers - Integrate existing AI services into your applications
- Configure authentication - Set up secure connections to third-party MCP servers
- Make model-specific requests - Use specialized features of different AI models
- Handle responses - Process and utilize the responses from various MCP servers
Overview
TRMX AI allows you to use existing MCP servers in your applications without having to build your own. This approach lets you leverage pre-built, specialized AI capabilities with minimal setup.
Available MCP Servers
Popular MCP Services
These production-ready MCP servers offer specialized AI capabilities:
- Claude: Advanced natural language understanding and reasoning
- Cursor: Specialized code completion and generation
- Windsurf: Document processing, analysis, and extraction
Getting Started
Prerequisites
Before connecting to external MCP servers, you'll need:
- The TRMX SDK installed (
npm install @trmx/sdk
) - API keys for the services you want to use
- Basic familiarity with async/await JavaScript patterns
Integration Example
Here's how to connect to and use various MCP servers:
// Import the TRMX client
const { TrmxClient } = require('@trmx/sdk');
// Initialize with your API key
const client = new TrmxClient({
apiKey: process.env.TRMX_API_KEY
});
// Use Claude MCP server
async function askClaude(question) {
const response = await client.connect('claude').ask(question);
return response.text;
}
// Use Cursor MCP server
async function generateCode(prompt) {
const response = await client.connect('cursor').generate({
prompt,
language: 'javascript'
});
return response.code;
}
// Use Windsurf for document processing
async function analyzeDocument(documentUrl) {
const response = await client.connect('windsurf').analyze({
documentUrl,
tasks: ['summarize', 'extract_entities']
});
return response.results;
}
Authentication
Each MCP server may require specific authentication methods. You'll typically need to:
- Register on the service's platform
- Generate an API key
- Include the key in your connection configuration
Security Best Practices
When working with external MCP servers:
- Store API keys in environment variables, never in code
- Use proper error handling for service disruptions
- Implement rate limiting to avoid unexpected costs
- Consider fallback options if a service is unavailable
Next Steps
Continue Learning
Now that you know how to use existing MCP servers:
- Test your MCP integrations - Ensure your connections work properly
- Explore advanced features - Use streaming responses for real-time applications
- Set up authentication - Implement secure user authentication