In this article, we’ll dive into the world of Custom MCP Catalogs and Profiles, two game-changing features for managing AI tools at scale. These capabilities help teams curate, share, and run Model Context Protocol (MCP) servers efficiently. Below, we answer common questions about how they work, why they matter, and how to get started.
Jump to: What are Custom MCP Catalogs and Profiles? | Why are Custom Catalogs important for enterprises? | How do I create a custom MCP catalog with Docker? | Step-by-step: building and sharing a custom catalog | How do MCP Profiles improve developer workflows? | What types of MCP servers can be included in a catalog?
What are Custom MCP Catalogs and Profiles?
Custom MCP Catalogs are curated collections of approved MCP servers that organizations can publish and distribute internally. Instead of each developer hunting for servers across the open internet, catalogs provide a centralized, trusted source for AI tooling. Profiles, on the other hand, are portable, named groupings of MCP servers that individual developers can define. They allow you to bundle specific servers together, run them easily, and share configurations across projects or teams. Think of catalogs as the enterprise-level library, while profiles are the developer’s personal toolkit. Both work together to streamline discovery, distribution, and management of MCP resources, reducing friction and ensuring consistency.
Why are Custom Catalogs important for enterprise MCP adoption?
As organizations adopt MCP, a recurring need emerges: teams must curate a trusted list of MCP servers, including internally built ones. Without a central catalog, every employee searches independently, risking security and compatibility issues. Custom Catalogs solve this by allowing companies to publish approved server lists. Developers can then discover and use only vetted tools. This brings flexibility, control, and trust into one experience. It also simplifies onboarding — new team members instantly know which servers to use. By integrating with Docker’s existing MCP Catalog, community sources, and custom servers, organizations can enforce standards without stifling innovation.
How do I create a custom MCP catalog with Docker?
Creating a custom MCP catalog with Docker involves several steps. First, you need MCP servers — either from the Docker MCP Catalog, community sources, or your own custom ones built as Docker images. For each server, you create a metadata file (e.g., a YAML file describing the server’s name, description, image location, and type). Then you combine these metadata files into a catalog that can be published. Docker Desktop provides a user interface to import catalogs, while the CLI allows full control. The catalog can reference servers from multiple sources, giving you flexibility. For example, you might include a popular database server from the official catalog alongside a proprietary analytics server your team built.
Step-by-step: building and sharing a custom catalog
Let’s walk through a concrete example. First, create a custom MCP server — say, a “roll-dice” server (a simple tool for rolling dice). Build it as a Docker image and push it to Docker Hub. Then create a YAML metadata file (e.g., mcp-dice.yaml) that defines the server: name, title, type, image, and description. Next, compile a catalog that includes this server along with any others you want, such as servers from Docker’s official MCP Catalog. You can do this via CLI commands or through Docker Desktop’s import feature. Once the catalog is ready, share it with your team or publish it to a registry. Developers can then load your catalog and immediately access the approved servers. For detailed CLI examples, refer to the Docker documentation.
How do MCP Profiles improve developer workflows?
MCP Profiles serve as portable, named groupings of MCP servers. They let developers define a set of servers needed for a specific project or task — for instance, a “data-analysis” profile that includes a database server, a compute server, and an NLP server. These profiles can be saved, shared, and run across different environments. This solves real-world problems: no more manually configuring servers every time you switch projects, no more copying configurations between machines. Profiles also provide a foundation for future enhancements, like versioning and dependency management. By making setups repeatable and shareable, profiles accelerate development and reduce errors.
What types of MCP servers can be included in a catalog?
Custom MCP catalogs are designed to be inclusive. You can include servers from the official Docker MCP Catalog, which are pre-vetted and ready to use. You can also add community‑developed servers from public repositories. Most importantly, you can include your own internally built MCP servers — for example, a custom search tool or a private data connector. The only requirement is that each server can be described with metadata (name, image location, etc.) and, if using Docker, is packaged as a Docker image. This flexibility allows organizations to mix trusted external tools with proprietary ones, all in one catalog. The result is a unified discovery experience where developers see exactly the servers they are allowed to use.