Docker MCP Enterprise Gateway governs how MCP-compatible clients access approved servers and tools. It applies identity-aware policy, supplies credentials from an approved secret store, and records tool-call decisions for enterprise teams.
Docker MCP Enterprise Gateway is an enterprise gateway between MCP-speaking clients and MCP servers. It authenticates the user, determines which servers and tools are available to their groups, evaluates policy for each call, supplies credentials from an approved secret store, routes the request, and records the decision.
It is designed to centralize governance across remote services, containerized servers, and internal servers built by an organization. Administrators manage approved servers and tool access centrally, while developers connect their client to a gateway endpoint.
The gateway resolves access using the user’s identity and groups, then evaluates rules for the specific call. Rules can allow or deny access by server, tool, transport, and call; unapproved servers are denied by default.
Administrators add and publish servers for selected groups and expose only the tools those groups need. Developers connect their client once, and catalog or policy changes take effect without individual client reconfiguration.
Credentials are retrieved from the organization’s approved secret store when a tool runs, rather than being distributed through client configuration files.
Each policy evaluation produces a structured event associated with the user and agent. Events can be streamed to the organization’s SIEM pipeline.
The gateway handles remote services over HTTP or SSE, containerized servers, and internally built servers. Containerized servers run in isolated containers. Deployment options listed include a Docker-operated instance in a customer cloud account or VPC and an air-gapped appliance on private Kubernetes.
Agents can discover and invoke tools as needed instead of loading every tool definition up front, reducing the amount of tool catalog context loaded for each request.
A platform team can replace separate MCP configurations with a centrally approved catalog and group-based access, allowing developers using different agent harnesses to follow the same governance rules.
Teams can connect internally built MCP servers and have the gateway provide credentials at call time, avoiding long-lived credentials in client configuration files.
Security teams can use identity-linked events for individual tool-call evaluations and stream those records into an existing SIEM pipeline.
Organizations that require an air-gapped deployment can run the appliance on private Kubernetes, loading catalogs, server images, and policy from within the network without an outbound connection.
The gateway supports clients that speak MCP. The product page names Claude Code, Cursor, VS Code, Codex, ChatGPT Enterprise, and agents built in-house.
It can connect to remote services over HTTP or SSE, containerized servers, and internal servers built by an organization. Containerized servers run in isolated containers.
Credentials remain in the organization’s approved secret store. The gateway supplies them when a tool runs, rather than distributing long-lived credentials through client configuration files.
Yes. The air-gapped appliance runs on private Kubernetes and loads catalogs, server images, and policy from inside the network.
Docker Sandboxes governs the agent’s execution environment, including its network, files, and MCP servers. The gateway governs which servers and tools a client can call. Either can be used independently, or both can operate under central policy.
glama.ai
Glama is a registry and infrastructure platform for discovering, evaluating, and connecting to Model Context Protocol (MCP) servers, connectors, clients, and tools. It supports browsing open-source and remote MCP entries, with hosted access and gateway features described on the site.
0mcp.io
0mcp converts selected operations from an OpenAPI-described API into a hosted Model Context Protocol (MCP) server. It helps SaaS, platform, and internal teams make approved API actions available to MCP-compatible AI clients without implementing and operating the protocol layer themselves.
app.dezycro.ai
Dezycro is an operations layer for AI coding agents that provides product context, steers work, verifies behavior against requirements, and records decisions. It is designed for developers, product teams, and organizations that want to use agents without giving Dezycro access to source code.
theagentrouter.ai
Agent Router is an open-source Go project for managing unified access to generative AI services through Envoy Gateway. It is intended for teams that want a gateway-based control point for AI service traffic and related agent or MCP workloads.
gitmcp.io
GitMCP creates a remote Model Context Protocol (MCP) server for public GitHub repositories and GitHub Pages sites. It helps MCP-compatible AI assistants access repository documentation and code context for more relevant coding responses.
modelcontextprotocol.io
The official TypeScript SDK for building Model Context Protocol servers and clients. It provides a repository-organized foundation for MCP development, with separate client, server, core, middleware, documentation, and example areas.
