Ctrlk
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Hosting MCP servers and practical examples

How MCP server hosting works

AIsuru does not manage the hosting of MCP servers. The platform only connects to the MCP servers you configure — it doesn't host them directly. This gives you full flexibility in choosing where and how to run your servers.

Two hosting models

When working with MCP servers, you'll mainly encounter two models:

1. Third-party managed MCP servers

Some services offer MCP servers that are already hosted and managed by the platform itself. In this case:

  • the MCP server is already running on their systems;

  • you simply receive access credentials (API key, URL, etc.);

  • you connect directly to their service.

Practical example: Monday

  • Monday hosts and manages the MCP server for you;

  • they give you an API key and an endpoint;

  • you enter these parameters in AIsuru and you're ready to go.

2. Self-hosted MCP servers

Other MCP servers require you to install and manage them yourself. In this case:

  • you download the MCP server software;

  • you install it on an infrastructure of your choice;

  • you configure it to suit your needs;

  • you obtain the credentials to connect.

Practical example: AutoCAD MCP Server

  • you need to install the MCP server in your infrastructure;

  • you configure access to AutoCAD;

  • you get the URL and credentials for your server;

  • you enter these details in AIsuru to connect.

Hosting options for self-hosted servers

When you need to host an MCP server yourself, you have several options:

  • cloud hosting;

  • virtual machines;

  • local installation;

  • Docker and containerization.

Cloud hosting

Run the server on a cloud provider such as:

  • AWS (Amazon Web Services);

  • Azure (Microsoft);

  • Google Cloud Platform;

  • DigitalOcean;

  • Linode.

Advantages: high availability, scalability, accessible from anywhere

Virtual machine (VPS)

Use a dedicated Virtual Private Server:

  • Hetzner;

  • OVH;

  • Contabo.

Advantages: greater control, predictable costs

Local installation

Run the server directly on your computer or on a company server:

  • local workstation;

  • on-premises company server;

  • Docker container on your machine.

Advantages: maximum control, no external hosting costs, data stays within your network

Docker and containerization

Use Docker to run the server in containers:

  • on a local machine;

  • on a cloud server;

  • on a Kubernetes cluster.

Advantages: isolation, portability, easy deployment

How AIsuru connects to servers

Regardless of where your MCP server is hosted, AIsuru:

  1. receives the connection parameters from you (URL, credentials, etc.);

  2. connects to the server using these parameters;

  3. communicates with the server to execute the required operations.

Important note: AIsuru doesn't store MCP server data — only the information needed to establish the connection.

Practical examples:

In this section we'll look at some practical examples of MCPs available in the platform and how they could be self-hosted or run in Docker, specifically MongoDB.

Hosting options:

  • Managed cloud: MongoDB Atlas (the server is hosted by MongoDB)

  • Local Docker: docker run -d -p 27017:27017 mongo

  • Direct installation: MongoDB installed on your system

  • Other cloud: AWS DocumentDB, Azure Cosmos DB

Local MongoDB MCP

Connection String:

With authentication:

Database Name: your choice (e.g., test, local)

MongoDB in Docker

  1. Start the container:

  2. Connection String:

  3. Database Name: whatever you prefer.

Last updated