Managing AI tooling across an enterprise is no small feat. With the explosion of Model Context Protocol (MCP) servers, teams need a way to standardize, distribute, and run these tools without chaos. That’s why Docker’s new Custom MCP Catalogs and Profiles are a game changer. They give organizations the power to curate trusted servers, while developers get portable configurations they can share and reuse. In this article, we break down the six most important things you need to know about these new capabilities—from what they are to how they transform your MCP workflows.
1. What Are Custom MCP Catalogs?
Custom MCP Catalogs let organizations curate and distribute approved collections of MCP servers. Instead of each team member hunting for servers across the open internet, admins can publish a single catalog that defines which servers are trusted and allowed. These catalogs can reference servers from Docker’s MCP Catalog, community sources, and internally built servers—all in one place. This centralizes discovery and ensures compliance. For example, a company might include fetch from Docker’s catalog and a custom server created by their AI team. The result: developers spend less time searching and more time building, while security teams sleep better knowing only vetted tools are in use.
2. What Are MCP Profiles?
MCP Profiles are portable, named groupings of MCP servers. They allow individual developers to build, run, and share their MCP tools and configurations across projects and teams. Think of them as a lightweight way to package your “MCP stack.” If you’re working on a project that needs a database server, a search tool, and a custom integration, you can define a profile that bundles all three. Then share that profile with teammates—no more copying config files or forgetting which server goes where. Profiles solve practical use cases today, like switching between dev and production setups, and lay the groundwork for even more advanced features in future releases.
3. Why Organizations Need Curated MCP Servers
As MCP adoption grows, so does the risk of using unverified or insecure servers. In uncontrolled environments, developers might grab any server they find, potentially introducing vulnerabilities or incompatible tools. Curated catalogs solve this by creating a trusted walled garden. They allow organizations to enforce which servers are allowed, ensure consistent versioning, and provide a single source of truth. This is especially critical for enterprises in regulated industries where every tool must be audited. The result is faster onboarding for new team members, reduced duplication of effort, and better governance—all without sacrificing developer flexibility.
4. How to Build a Custom MCP Server and Catalog
Building a custom MCP server is straightforward. Start by creating a standard MCP server that communicates over stdio (Docker provides a reference server called roll-dice). Package it as a Docker image and push it to Docker Hub. Then define a metadata file—say mcp-dice.yaml—that describes the server, including its image location. Finally, create a custom catalog that combines your server with others from the Docker MCP Catalog. The entire process can be done via the CLI. For example, you might run commands to add your server’s metadata, specify the catalog name, and publish it. This gives you full control over what’s included and how it’s versioned.
5. How to Import and Use Custom Catalogs in Docker Desktop
Once you’ve created a custom catalog, importing it into Docker Desktop is a breeze. With a few clicks (or commands), you can add the catalog to your local environment. Docker Desktop will then display all the approved servers from that catalog, making them available for use in your projects. This is great for developers who want the curated experience without leaving their favorite UI. For CLI power users, the same functionality is available through terminal commands. The catalog shows up alongside the default MCP Catalog, and you can easily toggle between them. This unified experience means less friction for teams adopting new tools.
6. The Future of MCP Profiles and Catalogs
These two capabilities are just the beginning. Profiles give us a foundation to expand into more sophisticated scenarios, like environment-specific profiles (staging vs. production), team-based sharing, or even dynamic profiles that auto-update when new servers are approved. Catalogs, meanwhile, can evolve to support hierarchical organizations, role-based access, and integration with CI/CD pipelines. Together, they represent a shift toward treating AI tooling with the same discipline as code dependencies. Organizations that adopt this early will gain a competitive edge in managing their MCP ecosystem efficiently and securely. The future is about making MCP servers as easy to manage as npm packages – and we’re well on our way.
Custom MCP Catalogs and Profiles are now generally available, and they solve real pain points for teams scaling their AI tooling. Whether you’re a platform engineer looking to enforce standards or a developer wanting to share your favorite tools, these features give you the control and portability you need. Start with a small catalog that includes your most critical servers, then experiment with profiles to bundle them for different projects. The result: faster development, better governance, and a happier team. Ready to dive in? Docker Desktop is the easiest place to get started.