AWS Documentation MCP Server
MCP setupDevOps and cloud
FreeMCP setupDevOps and cloud
Route your agent through Docker's MCP gateway so each tool server runs in its own isolated container with secrets kept out of env variables.
npx workspacetemplates add docker-mcp --tool claude-codeWrites 1 file to your project root
{
"mcpServers": {
"docker": {
"command": "docker",
"args": [
"mcp",
"gateway",
"run"
]
}
}
}npx workspacetemplates add docker-mcp --tool codexWrites 1 file to ~/.codex/
[mcp_servers.docker]
command = "docker"
args = ["mcp","gateway","run"]npx workspacetemplates add docker-mcp --tool cursorWrites 1 file to .cursor/
{
"mcpServers": {
"docker": {
"command": "docker",
"args": [
"mcp",
"gateway",
"run"
]
}
}
}npx workspacetemplates add docker-mcp --tool gemini-cliWrites 1 file to .gemini/
{
"mcpServers": {
"docker": {
"command": "docker",
"args": [
"mcp",
"gateway",
"run"
]
}
}
}npx workspacetemplates add docker-mcp --tool copilotWrites 1 file to .vscode/
{
"servers": {
"docker": {
"type": "stdio",
"command": "docker",
"args": [
"mcp",
"gateway",
"run"
]
}
}
}The Docker MCP gateway sits between your coding agent and a set of MCP servers. Instead of starting each server yourself, you pick them in a profile and the gateway runs the local ones in separate containers, with minimal access to your machine.
Your agent sees one server and the gateway handles the rest. You can also turn individual tools on or off per profile, which is a simple way to keep a risky tool out of reach. It needs Docker Desktop 4.59 or later with the MCP Toolkit enabled.
you › List the tools you can reach through the Docker gateway and say which server each one comes from.
agent › 12 tools from 3 servers in the default profile: fetch (1 tool), github (9 tools) and duckduckgo (2 tools). Each runs in its own container.MCP setupDevOps and cloud
FreeMCP setupDevOps and cloud
FreeMCP setupDevOps and cloud
FreeMCP setupDevOps and cloud
Free