What is an MCP Server?
New to MCP servers? Learn what they are, how they work, and why you might want to subscribe to one to extend your AI assistant's capabilities.
You're Viewing a Tool That Adds Capabilities to Your AI
You landed here because you clicked "What is this?" on a server info page. Here's what you need to know: this is a tool that extends what your AI assistant can do - like giving it access to live data, specialized APIs, or new features that go beyond its base capabilities.
This page will help you understand what MCP servers are, how they work, and whether subscribing to one makes sense for you.
What is mctx?
mctx is the hosting platform you're viewing this server on. When you see a .mctx.ai domain, it means a developer has chosen mctx to host their MCP server. mctx handles the infrastructure, payments, and distribution so developers can focus on building useful tools.
The server info page shows you what the server does, how much it costs, and how to subscribe.
What is an MCP Server?
Think of your AI assistant like a smartphone, and MCP servers like apps. Your phone is powerful, but it needs apps to check weather, navigate, or order food. Similarly, your AI needs MCP servers to access specialized data and perform specific tasks.
MCP servers add features to your AI tools. When you connect an MCP server to your AI assistant (like Claude Code, Cursor, or similar tools), the AI can use the server's capabilities to accomplish tasks it couldn't do alone.
Real example: You're working in Claude Code and ask "What's the weather in Tokyo?" Your AI assistant calls a weather MCP server's tool, retrieves the current conditions, and gives you the answer - all seamlessly in your conversation.
What Can MCP Servers Do for You?
MCP servers solve problems that AI models can't tackle on their own. Here's what they enable:
Access Live Data
Your AI can retrieve information that changes in real-time:
- Check weather forecasts for any city
- Look up current stock prices or cryptocurrency rates
- Pull the latest news headlines
- Search specialized databases or research papers
Interact with Services and APIs
Your AI can perform actions on your behalf:
- Create GitHub issues or review pull requests
- Manage cloud services (AWS, GCP, Azure)
- Control project management tools
- Send emails or notifications
Work with Files and Documents
Your AI can process and transform content:
- Search through files and folders
- Convert between document formats
- Process images or videos
- Generate reports from data
Perform Specialized Computations
Your AI can handle complex analysis:
- Execute mathematical calculations
- Run simulations or data analysis
- Convert between technical formats
- Process data pipelines
All of this happens automatically - you don't manually call APIs or copy-paste data. The AI handles the orchestration behind the scenes.
How Do MCP Servers Work?
The interaction is seamless:
- You ask your AI a question or give it a task
- Your AI recognizes it needs external data or capabilities
- The AI calls the appropriate MCP server with the right parameters
- The MCP server processes the request and returns the result
- Your AI incorporates the result into its response to you
You stay in your normal workflow. The complexity is hidden.
Is This Right for Me?
MCP servers are useful if:
- You use AI tools like Claude Code, Cursor, or similar assistants regularly
- You've wished your AI could access live data, call APIs, or use specialized tools
- You want your AI to handle tasks that currently require you to manually copy-paste data or switch between tools
MCP servers might not be worth it if:
- You rarely use AI assistants
- Your work doesn't require real-time data or API access
- You prefer doing tasks manually rather than delegating to AI
What's the Catch?
Do I need coding skills?
No. Installation is simple: copy and paste one command into your AI tool's settings. If you can follow installation instructions, you can use MCP servers.
What if I don't use it much?
Pricing is usage-based, so you pay only for what you use. If you make just a few requests, your cost will be minimal - often pennies for testing.
Can I cancel anytime?
Yes. You can cancel your subscription anytime. When you cancel, your subscription is canceled immediately and you will not be charged again. Access to the server typically ends within minutes of cancellation. No cancellation fees.
What if the server doesn't work?
Servers hosted on mctx are versioned and monitored. If you run into issues:
- Check the FAQ for common troubleshooting steps
- Review your authentication and subscription status in your dashboard
- Contact the developer (support link available on the server info page)
Is my payment information secure?
Payment is handled by Stripe, one of the largest payment processors in the world. mctx never stores your credit card details.
Why Subscribe to an MCP Server?
When you subscribe to a server on mctx, you get:
Access to specialized capabilities - Use the server's tools through any compatible AI client
Locked-in pricing - Your rate never changes while subscribed, even if prices increase later. The price you see when you subscribe is locked in for as long as you maintain that subscription.
Version control - Choose which version to use, update when you're ready
Automatic sign-in - Authentication is handled by your AI client automatically. No manual token management.
Usage-based billing - Pay only for what you use. No charges if you don't use the server.
How to Subscribe
Step 1: Review the Server
On the server info page, check:
- Name and description - what does it do?
- Available tools - what capabilities does it provide?
- Price - cost per 1,000 requests
- Versions - which versions are available
Step 2: Complete Checkout (Under 5 Minutes)
- Click the Subscribe button
- Sign in with your email
- Enter your payment details
- Complete the checkout
You'll be redirected to your subscriptions dashboard.
Step 3: Install the Server
Installation is simple: copy and paste one command into your AI tool's settings.
After subscribing, expand the Installation Instructions section on your subscription card. You'll see commands for different AI clients with the exact server details pre-filled:
Claude Code:
claude mcp add --transport http {slug} https://{slug}.mctx.ai/v{version}Cursor:
Don't worry - just copy this exactly as shown:
{
"mcpServers": {
"{slug}": {
"url": "https://{slug}.mctx.ai/v{version}"
}
}
}Codex:
codex mcp add {slug} --url https://{slug}.mctx.ai/v{version}The installation instructions show the exact commands with the correct server slug and version.
Step 4: Authenticate
When you first use the server, your AI client will:
- Open a browser window for authentication
- Prompt you to sign in with your mctx account
- Store the access token securely
This happens automatically the first time you use the server.
Step 5: Start Using
Your AI assistant can now use the server's tools. For example:
"Use the weather tool to get the current weather in San Francisco"
Your AI will call the MCP server and return the result.
Understanding Pricing
How Billing Works
mctx uses usage-based billing:
- Servers charge per request (typically per 1,000 requests)
- You're billed monthly based on how many requests you make
- No charges if you don't use the server
Example:
- Server charges $0.50 per 1,000 requests
- You make 2,500 requests in a month
- You're charged $1.25 that month
Locked-In Pricing
The price you see when you subscribe is locked in for as long as you maintain that subscription. Your rate never changes while you're subscribed, even if the developer raises prices later.
This protects you from unexpected price increases while subscribed. However, if you unsubscribe and resubscribe later, you'll pay the current price at that time - not your original locked-in rate. Early subscribers who stay subscribed always pay less.
Can I try before I subscribe?
mctx doesn't offer free trials. Servers are paid to ensure developers are compensated and servers remain maintained.
However, usage-based billing means test requests cost pennies. If you're exploring whether a server fits your workflow, a few test requests will cost far less than a dollar.
What AI Tools Work with mctx?
Any AI client that supports MCP and automatic sign-in works with mctx servers. Popular clients include:
- Claude Code
- Cursor
- Codex
- Claude Desktop (with compatible plugins)
- Custom integrations using the MCP SDK
What Happens If I Cancel?
- When you cancel, your subscription is canceled immediately and you will not be charged again. Access to the server typically ends within minutes of cancellation.
- No further charges after you unsubscribe
- You can resubscribe anytime at the current price (not your old locked-in price)
If you resubscribe later, you'll pay the new rate - not the original price you had locked in.
Want to Build Your Own MCP Server?
Developers can host MCP servers on mctx and earn revenue from subscriptions. You keep 80%, mctx takes 20% and handles the rest.
Learn more: MCP Basics and Server Requirements
Next Steps
Ready to subscribe? Head back to the server info page and click Subscribe - you'll have access in under 5 minutes.
Want to learn more about using MCP servers? See the FAQ for common questions.
See something wrong? Report it or suggest an improvement — your feedback helps make these docs better.