Guides
Start typing to search…

What is hosted CustomGPT.ai MCP server?

This step-by-step guide explains what the hosted CustomGPT.ai MCP server is, how it works, and how you can use it to connect your agents with other AI systems through the Model Context Protocol (MCP).

What is MCP?

Model Context Protocol (MCP) is an open standard that lets AI models use external tools and data through a single, consistent interface.

In simple terms, MCP:

  • Gives AI models a standard way to call tools and access data.
  • Uses an MCP server to expose tools.
  • Uses an MCP client (such as Claude, ChatGPT, or another MCP-compatible app) to connect to that server and run those tools.

Instead of building many custom integrations, MCP lets you “connect once and reuse” with any client that supports the protocol.

What is the hosted CustomGPT.ai MCP server?

Every CustomGPT agent includes its own hosted MCP (Model Context Protocol) server. This server makes it easy to connect your agent to other AI systems—such as Claude, ChatGPT, or any MCP-compatible client—so they can use your agent’s tools and knowledge base through a secure, permission-controlled interface.

Hosted MCP server for every agent

CustomGPT automatically hosts an MCP server for every agent you create, at no cost.

This means:

  • Every agent has its own MCP server.
  • Each MCP server is isolated, so access is scoped to that specific agent.
  • You do not need to deploy or manage any extra infrastructure.
  • You can instantly connect any external AI client to any agent.

You can take an agent’s MCP server link and use it in any MCP-compatible client to let that client interact with the agent.

What can the hosted MCP server do?

Through the hosted MCP server, an external AI can use tools that operate on your CustomGPT agent. The exact capabilities depend on which permissions you enable.

Examples of what an external AI can do:

  • Query the knowledge base: Ask questions and get answers from your agent’s knowledge.
  • Work with conversations: Send messages, read past messages, or fetch full conversations.
  • Access or manage documents: Read knowledge documents and sources, or update and delete them.
  • Read or update settings: View or change agent and personalization settings.
  • View analytics: Retrieve analytics and deployment stats.

For a complete list of permissions and what they allow, see: MCP server permissions.

Connect your agent to other AI systems

Using the hosted MCP server link, you can connect your agent to:

This allows those clients to call your agent’s tools directly.

For guidance on deploying your agent via MCP, see: Deploy using MCP server.


Updated 1 day ago