Quick Start Guide

This guide will walk you through creating your first MCP tool from scratch. By the end, you'll have a working tool that an AI assistant can use. The entire process takes just a few minutes, and we'll explain what each step does and why it matters.

Step 1: Create an Organization

An organization is the top-level container for everything you'll build on MCP Functions. Think of it as your company's account or workspace on the platform. All your tools, team members, and billing are associated with your organization.

Why you need an organization: Organizations allow you to manage team access, control billing, and organize all your MCP tools in one place. If you're working solo, you'll still create an organization - it's just for you. If you're part of a team, the organization is where everyone collaborates.

Here's how to create one:

• Sign up or log in - Visit the MCP Functions website and either create a new account or log in with your existing credentials. The signup process is straightforward and only takes a minute.

• Click "Create Organization" - Once you're logged in, you'll see a dashboard. Look for the "Create Organization" button, which is typically prominently displayed for new users.

• Enter your organization details - You'll need to provide:

• **Organization name** - This is the display name, like "Acme Corporation" or "My Company". This is what you and your team will see. - **Organization slug** - This is a URL-friendly version of your name (like "acme-corp"). It's used in URLs and API endpoints. The system will suggest one based on your name, but you can customize it. Once set, it can't be changed, so choose carefully.

- **Invite team members (optional)** - If you're working with others, you can invite them now or later. Team members can be given different roles (Owner, Admin, or Member) that control what they can do in the organization.

What happens next: Once your organization is created, you'll be taken to the organization dashboard. This is your home base where you'll see all your workspaces, MCP servers, and tools.

Step 2: Create a Workspace

Workspaces are isolated environments within your organization where you group related projects and tools. Think of a workspace as a folder or project directory - it helps you organize your tools logically.

Why you need workspaces: As you build more tools, you'll want to organize them. You might have one workspace for "Customer Support Tools", another for "Sales Automation", and another for "Internal Operations". Workspaces also help you control access - you can give different team members access to different workspaces.

When to create multiple workspaces: You might create separate workspaces for: - Different projects or products - Different teams or departments - Different environments (development, staging, production) - Different clients (if you're building tools for multiple companies)

For your first tool, you can create a simple workspace like "My First Tools" or "Testing". Here's how:

• Navigate to your organization - From the organization dashboard, you'll see a list of workspaces (which will be empty at first).

• Click "Create Workspace" - This button is usually at the top of the workspaces list or in a prominent location on the dashboard.

• Enter workspace details:

• **Name** - Give your workspace a descriptive name like "Customer Support Tools" or "My First Workspace" - **Description** - Optionally describe what this workspace is for. This helps you and your team remember its purpose later.

- **Configure settings (optional)** - For your first workspace, you can use the default settings. As you get more advanced, you might configure things like rate limits, execution timeouts, or environment variables.

What happens next: Once created, you'll be taken into the workspace view, where you can see all the MCP servers and tools within this workspace.

Step 3: Create an MCP Server

An MCP Server is a container that groups related tools together and exposes them via the MCP protocol. This is what AI assistants actually connect to when they want to use your tools.

Why you need an MCP Server: Think of an MCP server as a "toolbox" that holds multiple related tools. When an AI assistant connects to your MCP server, it can see all the tools inside and use any of them. For example, you might have an MCP server called "Customer Data Tools" that contains tools for looking up customers, updating customer records, and creating support tickets.

How AI assistants use MCP Servers: When you configure an AI assistant (like ChatGPT or Claude) to use your tools, you give it the connection details for your MCP server. The AI then connects to that server, discovers what tools are available, and can use them whenever needed.

Here's how to create one:

• Go to your workspace - From the workspace view, you'll see a list of MCP servers (empty at first).

• Click "Create MCP Server" - This will open a form to create a new server.

• Enter server details:

• **Name** - Give your server a descriptive name like "Customer Data Tools" or "Weather Services" - **Description** - Explain what this server does and what kinds of tools it contains. This helps you and your team understand its purpose.

- **Set up authentication** - MCP servers require an API key for security. This ensures that only authorized AI assistants can connect to your tools. You can either:

• Create a new API key specifically for this server - Use an existing API key if you have one The API key is what you'll provide to AI assistants when configuring them to use your tools. Keep it secure - treat it like a password.

What happens next: Once your MCP server is created, you'll see it in your workspace. It will be empty at first - no tools yet. That's what we'll create in the next step.

Step 4: Create Your First Tool

Now for the exciting part - creating your first tool! A tool is an individual function that performs a specific task. This is what actually does the work when an AI assistant needs to perform an action.

What a tool does: A tool is like a function in programming - it takes some inputs (parameters), performs some work, and returns a result. For example, a "Get Weather" tool might take a city name as input and return the current weather for that city.

How tools work with AI: When a user asks an AI assistant something like "What's the weather in New York?", the AI recognizes that it needs to use a tool to get this information. It calls your "Get Weather" tool with "New York" as the parameter, receives the weather data, and presents it to the user.

MCP Functions makes creating tools incredibly easy with AI-powered generation. Here's how:

• Open your MCP Server - Click on the MCP server you just created. You'll see an empty list of tools with a "Create Tool" button.

• Click "Create Tool" - This opens the tool creation interface.

• Choose creation method - You'll have two options:

• **AI-Powered Creation (Recommended)** - Describe what you want in plain English, and our AI generates the code - **Manual Creation** - Write the code yourself if you prefer full control For your first tool, we recommend using AI-powered creation to see how easy it is.

- **Describe your tool** - In the text box, describe what you want the tool to do. Be specific! For example:

• Good: "Create a tool that fetches current weather data for any city using the OpenWeatherMap API. It should take a city name as input and return temperature, humidity, and weather conditions." - Less helpful: "weather tool" (too vague) The more detail you provide, the better the AI can generate exactly what you need.

- **Review the generated code** - The AI will generate complete, production-ready code for your tool. You'll see:

• The tool's name and description (what the AI assistant will see) - Input parameters (what data the tool needs) - The actual code that performs the work You can review this code, make edits if needed, or regenerate if it's not quite right.

- **Test the tool** - Before deploying, you can test your tool right in the interface. Enter some sample inputs and see what it returns. This helps you catch any issues before the tool goes live. - **Save and deploy** - Once you're happy with the tool, click "Save" or "Deploy". The tool is immediately live and available for AI assistants to use! ### Example: Creating a Simple Tool

Let's say you want to create a tool that calculates the square of a number. Here's what you might type in the AI creation interface:

Create a tool that calculates the square of any number. It should take a number as input and return the square of that number. The tool should be called "calculate_square" and should validate that the input is actually a number.

The AI will generate code that: - Defines a tool called "calculate_square" - Accepts a number parameter - Validates the input is a number - Calculates and returns the square - Handles errors gracefully

Once deployed, an AI assistant can use this tool whenever a user asks something like "What's 15 squared?" - the AI will call your tool with 15, get back 225, and tell the user the answer.

💡 Tip: You can also write tool code manually if you prefer full control or want to implement complex logic. The manual creation option gives you a code editor where you can write JavaScript or Python code directly. This is useful when you need to integrate with specific APIs, implement complex business logic, or work with existing code libraries.

Step 5: Connect Your Tool to an AI Assistant

Now that your tool is created and deployed, you need to connect it to an AI assistant so it can actually be used. The exact steps depend on which AI assistant you're using, but the general process is similar.

What you'll need: To connect your MCP server to an AI assistant, you'll need: - The MCP server's connection URL (provided by MCP Functions) - Your API key (the one you created when setting up the MCP server)

How to find this information: In your MCP server settings, you'll find a section showing the connection details. This includes the URL that AI assistants should connect to and instructions for setting up the connection.

For ChatGPT/Claude: Each AI platform has its own way of configuring MCP servers. Generally, you'll: - Go to the AI assistant's settings or configuration - Find the section for "MCP Servers" or "Custom Tools" - Add a new server connection - Enter the connection URL and API key - Save the configuration

Testing the connection: Once connected, you can test it by asking the AI assistant to list available tools. It should show your tool! Then try using it - ask the AI a question that would require your tool, and watch it work.

What's Next?

Congratulations! You've created your first MCP tool. Here are some next steps:

• Create more tools - Build tools for different purposes and add them to your MCP server

• Explore AI-powered creation - Try creating more complex tools and see how the AI handles them

• Learn about manual creation - If you're comfortable with code, try writing tools manually for more control

• Organize your tools - Create additional workspaces and MCP servers as your needs grow

• Invite your team - Share your organization with colleagues so they can use and create tools

As you build more tools, you'll discover the power of MCP Functions - the ability to extend AI assistants with custom capabilities that are specific to your business needs, all without managing any infrastructure.