Deployment

Deployment Options

caddy-mcp supports local stdio, a loopback-only development listener, a least-privilege stdio container, and a remote authenticated HTTPS boundary. Provider endpoint, credential, selector, identity, and trust material are supplied at runtime through AgentConfig; none is stored in this repository.

Installed stdio process

{
  "mcpServers": {
    "caddy": {
      "command": "caddy-mcp",
      "args": [],
      "env": {"MCP_TOOL_MODE": "intent"}
    }
  }
}

Loopback development listener

caddy-mcp --transport streamable-http --host 127.0.0.1 --port 8000

Do not expose this listener beyond loopback. Network deployments require direct TLS or an explicitly trusted TLS-terminating ingress, configured authentication, exact MCP_ALLOWED_HOSTS, and an exact trusted-proxy CIDR policy.

Least-privilege local container

docker run -i --rm \
  --read-only \
  --cap-drop=ALL \
  --security-opt=no-new-privileges \
  --pids-limit=256 \
  --tmpfs /tmp:rw,noexec,nosuid,nodev,size=64m \
  -e TRANSPORT=stdio \
  registry.example.invalid/caddy-mcp@sha256:<digest> caddy-mcp

The operator projects the selected AgentConfig profile into the process at runtime; the image remains immutable and contains no environment connection profile.

Remote authenticated HTTPS endpoint

{
  "mcpServers": {
    "caddy": {"url": "https://service.example.invalid/mcp"}
  }
}

Store the real remote URL, outbound identity reference, and TLS-profile reference in AgentConfig, not in MCP client JSON or documentation.

This page covers running caddy-mcp as a long-lived server: the transports, a Docker Compose stack, the optional A2A agent server, putting it behind a Caddy reverse proxy, and giving it a DNS name with Technitium. To provision the Caddy server it connects to, see Backing Platform.

caddy-mcp ships two console scripts: caddy-mcp (the MCP tool surface) and caddy-agent (a Pydantic-AI A2A agent that consumes those tools). Deploy the MCP server on its own for tool access, or pair it with the agent server for a conversational interface.

Run the MCP server

The transport is selected with --transport (or the TRANSPORT env var):

stdio (default)streamable-httpsse

caddy-mcp

For IDE / desktop MCP clients that launch the server as a subprocess.

caddy-mcp --transport streamable-http --host 0.0.0.0 --port 8000

A network server with a /health endpoint and /mcp route.

caddy-mcp --transport sse --host 0.0.0.0 --port 8000

Health check (HTTP transports):

curl -s http://localhost:8000/health        # {"status":"OK"}

Configuration (environment)

caddy-mcp is configured entirely from the environment. The required set:

Var	Default	Meaning
CADDY_URL	http://localhost:2019	Caddy Admin API URL endpoint
CADDY_TOKEN	(unset)	Optional bearer token if the Admin API is secured

Plus HOST / PORT / TRANSPORT for HTTP transports, and CONFIGTOOL (default True) to register the configuration tool set. A template is provided in .env.example — copy it to .env and fill in your values.

Docker Compose

The repo ships docker/mcp.compose.yml. It reads a sibling .env and publishes the HTTP server on :8000:

services:
  caddy-mcp:
    image: example/caddy-mcp@sha256:<digest>
    container_name: caddy-mcp
    hostname: caddy-mcp
    restart: always
    env_file:
      - .env
    environment:
      - PYTHONUNBUFFERED=1
      - HOST=0.0.0.0
      - PORT=8000
      - TRANSPORT=streamable-http
      - CADDY_URL
      - CADDY_TOKEN
    ports:
      - "8000:8000"
    healthcheck:
      test: ["CMD", "python3", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/health')"]
      interval: 30s
      timeout: 10s
      retries: 3

cp .env.example .env          # then edit CADDY_URL / CADDY_TOKEN
docker compose -f docker/mcp.compose.yml up -d
docker compose -f docker/mcp.compose.yml logs -f

Agent server

caddy-mcp also ships an A2A agent server via the caddy-agent console script. The agent connects to a running MCP server (its tool source) and exposes a conversational Pydantic-AI interface over HTTP.

# Start the agent; point it at the MCP server's HTTP endpoint
caddy-agent --mcp-url http://caddy-mcp:8000/mcp --host 0.0.0.0 --port 9000

The agent reads its tool registration from mcp_config.json (bundled in the package) or from the --mcp-url of a remote MCP server. Provide a model provider with --provider / --model-id (or the corresponding environment variables). A Compose service for the agent mirrors the MCP service, wiring MCP_URL at its own published port (for example :9000):

# docker/agent.compose.yml
services:
  caddy-agent:
    image: example/caddy-mcp@sha256:<digest>
    container_name: caddy-agent
    hostname: caddy-agent
    restart: always
    entrypoint: ["caddy-agent"]
    env_file:
      - .env
    environment:
      - PYTHONUNBUFFERED=1
      - MCP_URL=http://caddy-mcp:8000/mcp
      - HOST=0.0.0.0
      - PORT=9000
    ports:
      - "9000:9000"
    depends_on:
      - caddy-mcp

Behind a Caddy reverse proxy

Expose the HTTP server on a hostname with automatic TLS. Add to your Caddyfile:

# Internal (self-signed) — homelab .example.invalid zone
caddy-mcp.example.invalid {
    tls internal
    reverse_proxy caddy-mcp:8000
}

# Public — automatic Let's Encrypt
caddy-mcp.example.com {
    reverse_proxy caddy-mcp:8000
}

Reload Caddy:

docker compose -f services/caddy/compose.yml exec caddy caddy reload --config /etc/caddy/Caddyfile

DNS with Technitium

Point the hostname at the host running Caddy. Via the Technitium API:

curl -s "http://technitium.example.invalid:5380/api/zones/records/add" \
  --data-urlencode "token=$TECHNITIUM_DNS_TOKEN" \
  --data-urlencode "domain=caddy-mcp.example.invalid" \
  --data-urlencode "zone=arpa" \
  --data-urlencode "type=A" \
  --data-urlencode "ipAddress=192.0.2.10" \
  --data-urlencode "ttl=3600"

…or add an A record caddy-mcp.example.invalid → <caddy-host-ip> in the Technitium web console (http://technitium.example.invalid:5380). The ecosystem technitium-dns-mcp automates this as a tool.

Register with an MCP client

Add to your client's mcp_config.json (multiplexer nickname cd):

{
  "mcpServers": {
    "caddy-mcp": {
      "command": "uv",
      "args": ["run", "caddy-mcp"],
      "env": {
        "CADDY_URL": "http://your-caddy:2019",
        "CADDY_TOKEN": ""
      }
    }
  }
}

For a remote HTTP server, point the client at http://caddy-mcp.example.invalid/mcp instead.