In addition to writing code, AI coding assistants can search documentation, engage in API calls, access databases, interact with repositories, and perform other actions on behalf of developers. For this, AI applications need a certain standard to talk to external applications and services.
This is where Model Context Protocol (MCP) is used.
MCP is an open protocol that standardizes the interaction of AI applications with external services and data sources. You can think of MCP as an interface between an AI application and the tools you already use. Instead of integrating each AI application individually into its environment, developers can use an MCP server that exposes the required tools to compatible MCP AI clients.
So, an AI coding assistant, for example, can retrieve current documentation, execute a database query, interact with GitHub, access an API, or use development tools, without the need to integrate with every workflow separately.
However, running MCP servers causes many of the issues associated with other software development tools – dependencies, configuration, credentials, security issues, and environment differences.
Docker MCP Catalog and Toolkit
Docker solves this issue with Docker MCP Catalog and Toolkit.
The MCP Catalog includes a set of pre-built and ready-to-use MCP servers available in containers. Instead of looking for a server, cloning it to the machine, installing dependencies, and configuring it, developers can easily discover MCP servers with Docker and run them in a container form.
The MCP Toolkit, built into Docker Desktop, allows developers to configure, manage, and run these containerized MCP servers. Developers can create profiles from MCP servers and connect these profiles to AI clients. Docker manages the runtime and dependencies, significantly reducing configuration issues.
This is especially handy if you already use Docker because you won’t need to install Node.js or Python dependencies on your machine for different MCP servers, but each of them will be run inside its own container.
One Gateway for Your AI Tools
Also, Docker provides an MCP Gateway as an additional layer between AI clients and MCP servers.
Without a gateway, developers may need to configure the same MCP server for every AI application they use. Docker’s MCP Gateway is a centralized server configuration layer, routing layer, authentication, credential management, and access control layer. If MCP Toolkit is enabled in Docker Desktop, Docker Gateway runs in the background automatically.
This means that developers can configure their MCP tools once and make them accessible for compatible AI clients.
Docker also provides MCP Enterprise Gateway for enterprises that require additional governance. The Enterprise Gateway allows controlling which servers and tools users can access, applying policies to tool calls, integrating with secret stores, and logging tool usage for audits.
Why Docker and MCP Work Well Together
One of the greatest achievements of Docker in software development was introducing portability and repeatability of applications across environments. MCP servers also benefited from this.
By containerizing MCP servers, developers get isolation and elimination of dependency and runtime issues. Also, Docker provides mechanisms for credential management and verification of MCP server image provenance.
This means that developers experimenting with AI-assisted workflows have fewer chances to spend time installing and configuring tools and more chances to use them.
MCP brings standardization of interaction between AI applications and developers’ tools. Docker makes MCP servers easy to discover, run, manage, and secure.
And if you already have Docker Desktop, getting started with MCP might be easier than you think.
Build Your Docker Skills
Want to strengthen your Docker fundamentals? Check out my Docker: Your First Project course on LinkedIn Learning. It’s part of the Docker Foundations Professional Certificate learning path and walks you through building a real project with Docker.