As AI-driven workloads become more complex, ensuring secure, scalable infrastructure is paramount. Docker's latest open-source project, the MCP Gateway, offers a robust solution for organizations moving agentic AI workloads from local development to production. By integrating seamlessly with Docker Compose and focusing on transparency, security, and community collaboration, the MCP Gateway sets a new standard for managing and safeguarding agent-based applications.
Building on a Strong Foundation
Docker's journey into agentic AI began with the MCP Toolkit, which standardized how developers define, run, and share agent-based workloads using Docker Compose. Community adoption has been rapid, with over a million pulls from the Docker MCP Catalog and a growing number of MCP servers in use.
The MCP Gateway expands on this foundation, providing a central, secure enforcement point between AI agents and external tools while remaining open source and community-driven.
Key Features and Workflow
- Discovery: Developers can easily browse available MCP servers using a simple CLI command:
docker mcp catalog show. The Docker MCP Catalog continues to evolve, with a PR-based process for contributing new servers and tools, encouraging community engagement and innovation.- Configuration: Security is baked in from the start. The MCP Gateway enables safe storage and injection of secrets (like API keys) and supports host-specific configurations through intuitive CLI interfaces. For example, setting a secret is as simple as:
docker mcp secret set 'brave.api_key=XXXXX'. Environment variables can be managed for different types of server runtimes, ensuring flexibility and safety across diverse workloads.- Running MCP Workloads: The gateway allows users to expose and manage multiple MCP server runtimes behind a single interface. Users can enable specific servers and even define custom gateway views that control which tools and transports are available to clients. This granularity makes it easy to tailor the environment for both development and production use cases.
- Security and Enforcement: A standout feature of the MCP Gateway is its ability to plug in generic interceptors. These can verify container image signatures, scan for secrets in payloads, and log calls, providing strong security guarantees. These security features are easily enabled via command-line flags, lowering the barrier for teams to adopt secure practices and maintain compliance.
Why the MCP Gateway Matters
Connecting AI agents to external tools can introduce security risks and operational complexity. The MCP Gateway addresses these challenges by consolidating server management and security enforcement into a single, consistent process. This gives developers confidence to scale their agentic applications from prototypes to production, knowing that governance and best practices are built into every step.
Open Source and Ready to Use
Docker has released the MCP Gateway as a fully open-source project, available in the latest Docker Desktop release and compatible with community editions. By sharing the project on GitHub, Docker invites developers, researchers, and teams to collaborate, extend, and innovate on this secure foundation for agentic AI workloads.
A Secure Path Forward for Agentic AI
The Docker MCP Gateway is more than just a tool—it's a strategic enabler for the next generation of AI-powered software. Whether you're building AI agents or supporting teams that do, the gateway offers a unified, secure, and community-driven approach to managing agentic workloads. Now is the perfect time to explore the MCP Gateway and help shape the future of secure AI infrastructure.
Docker MCP Gateway Provides Secure Agentic AI Deployment