Docker has released Docker Desktop 4.39. This version brings Docker Smart AI Agent with Model Context Protocol (MCP) and Kubernetes Integration, Docker Desktop CLI GA, and enhanced multi-platform support aiming to make it easier for users to manage multi-platform container images.
The Docker AI Agent, also known as “Ask Gordon”, has been updated; it now integrates with third-part services via Model Context Protocol (MCP), a new standard to connect AI agents and models to external data and tools. Developers can use it to generate GitHub projects, retrieve remote content, or even run AI agents as containers. To use it, users just need to type docker ai in the terminal or in the Docker Desktop AI Agent window, and the agent looks for a gordon-mcp.yml file in the working directory for a list of MCP servers to be used in that context.
Additionally, Docker AI now connects directly to the local Docker Engine and Kubernetes cluster to manage resources. It can list and clean up containers, images, and volumes, as well as interact with Kubernetes, such as listing namespaces, deploying services, and analyzing pod logs. These new integrations aim to expand what the AI assistant can do, reducing the need for manual commands and scripts in many common scenarios.
User experience has also been improved. The AI now supports multi-line prompts and copying output more easily within Docker Desktop. Response times have been reduced, and answers are more context-aware. According to Docker, ‘Ask Gordon’ now delivers responses faster and with deeper context about images, containers, and volumes, improving both performance and usefulness.
Docker Desktop CLI, previously in beta, is now generally available. First introduced as a beta in v4.37, the Docker Desktop CLI is a command-line interface tool that allows users to control Docker Desktop without clicking through the GUI. It allows developers to manage Docker Desktop directly from the command line, performing tasks such as starting, stopping, restarting, and checking the status of Docker Desktop. Additionally, the developers can also print logs and update to the latest version of Docker Desktop.
Docker Desktop 4.39 also extends the AI agent’s capability to support the containerization of the applications. Developers can now leverage Ask Gordon to generate Docker related assets for projects in the languages:
- JavaScript/TypeScript: using npm, pnpm, yarn, and bun.
- Go: using Go modules.
- Python: using pip, poetry, and uv.
- C#: using nuget.
To initiate the containerization process, developers just need to prompt Ask Gordon with a prompt like Can you containerize my application?. Ask Gordon will analyze the project, identify its structure, and generate essential Docker assets, including an optimized Dockerfile, Docker Compose file, .dockerignore file, and even a README with usage instructions.
(image taken by Docker blog)
Docker Desktop 4.39 has improved the support for multi-platform image management by introducing the --platform flag to the commands save, docker load, and docker history. Since containers today often need to run on multiple architectures such as deploying the same application to both x86_64 and ARM64 machines, this addition allows users to select specific architecture variants when exporting or importing images. It is possible for example for a user to export only the linux/arm64 version of an image with:
docker save --platform linux/arm64 -o custom-image.tar my-app:latest
and similarly import a specific architecture variant using:
docker load --platform linux/amd64 -i custom-image.tar
It is important to notice that Docker continues to support multi-platform builds using buildx.
More details about the news of the Docker Desktop can be found in the release notes.
