Streamlining Enterprise AI Tooling with Custom MCP Catalogs and Profiles

Organizations adopting the Model Context Protocol (MCP) for AI tooling face a common challenge: how to ensure teams use trusted, approved MCP servers while maintaining developer flexibility. Today, we’re excited to announce the general availability of Custom MCP Catalogs and MCP Profiles—two complementary features that transform how teams package, distribute, and manage MCP servers. Together, they provide a structured yet developer-friendly approach to enterprise MCP adoption.

Understanding Custom MCP Catalogs and Profiles

Before diving into the technical details, let’s clarify what these new primitives offer.

What Are Custom Catalogs?

Custom MCP Catalogs allow organizations to curate and distribute approved collections of MCP servers. Instead of each developer hunting for servers across the open internet, teams can publish a centralized catalog that defines exactly which servers—including internally built ones—are trusted and recommended. This brings flexibility, control, and trust into a single, authoritative source.

What Are MCP Profiles?

MCP Profiles are portable, named groupings of MCP servers. They enable individual developers to easily build, run, and share their MCP configurations across projects and teams. Profiles solve practical use cases today, such as switching between a local development setup and a production environment, while laying the groundwork for more advanced scenarios in the future.

Why Enterprises Need Curated MCP Server Management

As MCP adoption scales within organizations, a common need emerges: teams must be able to trust the servers they integrate. Without curation, developers may accidentally use unvetted or incompatible servers, leading to security risks, inconsistent behavior, and duplication of effort. Custom Catalogs address this by providing a single point of truth for approved MCP servers, whether they come from Docker’s official catalog, community sources, or in-house development.

Building and Sharing a Custom MCP Catalog

Let’s walk through the process of creating a custom catalog that combines servers from the Docker MCP Catalog with a server you built yourself. All steps can be executed via the CLI, while user‑facing features (like import) are also available in Docker Desktop.

In the examples below, we’ll use the Docker Hub ID roberthouse224 as a placeholder. You should substitute your own ID where appropriate (for example, when pushing an image).

Step 1: Creating and Dockerizing Your Own MCP Server

First, build a custom MCP server. For demonstration, we created a reference server called roll-dice (available on GitHub). It’s a standard MCP server that communicates over stdio and is packaged as a Docker image. After building, push the image to Docker Hub:

docker push roberthouse224/mcp-dice

Next, create a YAML metadata file (mcp-dice.yaml) that describes your server:

name: roll-dice
title: Roll Dice
type: server
image: roberthouse224/mcp-dice@latest
description: An MCP server that can roll dice

Step 2: Defining the Catalog Metadata

Now, create a catalog definition file (for example, my-company-catalog.yaml) that includes your custom server along with servers from the Docker MCP Catalog. The file references each server by its metadata location (a file or URL). A simplified example:

catalog:
  name: My Company's MCP Catalog
  servers:
    - file: ./mcp-dice.yaml
    - url: https://hub.docker.com/_/mcp/...  # Reference an approved public server
    - url: https://internal-repo/...        # Reference an internal server

Step 3: Publishing the Catalog

Publish the catalog so it can be accessed by your team. This could be as simple as hosting the YAML file on a shareable URL (e.g., an internal web server or a Git repository). Once published, any developer can reference the catalog in their Docker Desktop settings or CLI configuration.

Importing the Catalog into Docker Desktop

To import your custom catalog in Docker Desktop:

1. Open Docker Desktop and navigate to the MCP section.
2. Click Add Catalog and provide the URL or file path to your catalog YAML.
3. Docker Desktop will fetch the catalog and display the approved servers for quick selection.

Developers can now discover and use these servers directly, knowing they meet organizational standards.

Leveraging Profiles for Portable MCP Server Groupings

While catalogs address the “what” (which servers to use), Profiles tackle the “how to combine and reuse them”. An MCP Profile is a named configuration that groups several MCP servers together. For instance, you might define a “development” profile that includes a local SQLite server, a dice‑rolling server, and a custom API gateway, while a “production” profile substitutes the local SQLite with a cloud‑based database server.

Profiles are easily shareable via configuration files, enabling consistent setups across team members. They also simplify switching between contexts—no more manual reconfiguration of server endpoints.

Looking Ahead: The Future of MCP Management

Custom Catalogs and Profiles are just the beginning. We envision a future where administrators can set policies (e.g., mandatory approval workflows), and where Profiles can be dynamically composed based on project needs. The foundation we’re laying today makes it possible to build these advanced features while keeping the developer experience intuitive.

In summary, these two features give enterprises the governance they need without sacrificing the speed and creativity developers expect. By curating trusted MCP servers and enabling portable configurations, Custom Catalogs and Profiles accelerate enterprise MCP adoption.

Start building your custom MCP catalog today and explore Profiles to streamline your AI tooling workflow. For detailed CLI commands and configuration options, refer to the official Docker documentation.