For MCP Providers

This guide explains how to connect your existing MCP server to MCP Hive and make it available to consumers.

What is MCP Hive?

MCP Hive is a marketplace and gateway for Model Context Protocol (MCP) servers. As a provider, you can register your MCP server with MCP Hive to:

  • Reach a wider audience of AI-powered applications
  • Monetize your MCP tools and resources
  • Get usage analytics and performance metrics
  • Simplify credential management for your users

Prerequisites

Before registering your MCP server, ensure you have:

  • A working MCP server implementation
  • A publicly accessible endpoint (for remote MCPs)
  • An understanding of your MCP's tools, resources, and prompts

Registration Process

Step 1: Create an Account

  1. Navigate to the MCP Hive portal
  2. Click Get Started in the top navigation
  3. Fill in your email and password
  4. Verify your email address

Step 2: Access the Providers Section

  1. Log in to your MCP Hive account
  2. Click My MCP Servers in the sidebar navigation
  3. You'll see a list of your registered MCP servers (empty initially)

Step 3: Register Your MCP Server

  1. Click the Register Provider button
  2. Fill in the registration form:
FieldDescription
NameA display name for your MCP (e.g., "Weather API", "Code Analyzer")
DescriptionA clear description of what your MCP does
CategoriesComma-separated tags to help users discover your MCP
Deployment TypeChoose between Local or Remote
Endpoint URL(Remote only) The URL where your MCP server is hosted
Price per CallSet pricing in USD (use 0 for free)

Click Register to submit your MCP

Deployment Types

Remote MCP Servers

Remote MCPs are hosted on your own infrastructure and accessible via HTTPS. This is the recommended approach for production MCPs.

Requirements:

  • Publicly accessible HTTPS endpoint
  • Implements the MCP protocol correctly
  • Handles authentication via MCP Hive credentials

Local MCP Servers

Local MCPs run on MCP-Hive infrastructure.

Requirements:

  • Provide the npx command and arguments to start the server
  • Package your MCP for easy installation (e.g., npm package)

Technical Requirements

Your MCP server must:

  1. Implement the MCP Protocol - Follow the Model Context Protocol specification

  2. Define Tool Schemas - Each tool must have:

    • A unique name
    • A description
    • Input schema (JSON Schema format)
    • Required inputs clearly marked

  3. Handle Requests - Your server should properly handle:

    • tools/list - Return available tools
    • tools/call - Execute tool requests
    • resources/list - Return available resources (optional)
    • resources/read - Read resource content (optional)
    • prompts/list - Return available prompts (optional)
    • prompts/get - Get prompt content (optional)

Status Lifecycle

After registration, your MCP goes through a review process:

StatusDescription
DraftInitial state after registration
PendingAwaiting review by MCP Hive team
ApprovedApproved but not yet activated
ActiveLive on the marketplace, available to consumers
SuspendedTemporarily disabled (e.g., due to issues)

Managing Your MCPs

From the My MCP Servers section, you can:

  • View Details - See configuration and usage statistics
  • Edit - Update description, pricing, and other settings
  • Delete - Remove your MCP from MCP Hive

Best Practices

  1. Clear Descriptions - Write helpful descriptions so users understand what your MCP does

  2. Accurate Schemas - Define input schemas accurately to help AI models use your tools correctly

  3. Error Handling - Return meaningful error messages when requests fail

  4. Performance - Optimize response times for a better user experience

  5. Documentation - Consider providing external documentation for complex MCPs

Getting Help

If you need assistance: