Deployment¶

Deployment Options¶

ansible-tower-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": {
    "ansible-tower": {
      "command": "ansible-tower-mcp",
      "args": [],
      "env": {"MCP_TOOL_MODE": "intent"}
    }
  }
}

Loopback development listener¶

ansible-tower-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/ansible-tower-mcp@sha256:<digest> ansible-tower-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": {
    "ansible-tower": {"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 ansible-tower-mcp as a long-lived server: the transports, the companion A2A agent server, a Docker Compose stack, putting it behind a Caddy reverse proxy, and giving it a DNS name with Technitium.

ansible-tower-mcp ships two servers: an MCP server (console script ansible-tower-mcp) and an A2A agent server (console script ansible-tower-agent). The MCP server is a typed, deterministic tool surface; the agent server is a graph-routed Pydantic-AI agent that calls those tools over an MCP_URL.

Run the MCP server¶

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

stdio (default)streamable-httpsse

ansible-tower-mcp

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

ansible-tower-mcp --transport streamable-http --host 0.0.0.0 --port 8000

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

ansible-tower-mcp --transport sse --host 0.0.0.0 --port 8000

Health check (HTTP transports):

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

Configuration (environment)¶

ansible-tower-mcp is configured entirely from the environment. The connection to the Tower / AWX controller uses the following required set (provide a token, or a username / password pair, or an OAuth client id / secret pair):

Var	Default	Meaning
`ANSIBLE_BASE_URL`	none	Tower / AWX controller base URL (e.g. `https://tower.example.com`)
`ANSIBLE_USERNAME`	none	Controller user id
`ANSIBLE_PASSWORD`	none	Controller password
`ANSIBLE_TOKEN`	none	Pre-issued API token (bypasses username / password)
`ANSIBLE_CLIENT_ID`	none	OAuth application client id
`ANSIBLE_CLIENT_SECRET`	none	OAuth application client secret
`ANSIBLE_TOWER_TLS_PROFILE`	(unset)	Optional runtime TLS profile selector; verification is mandatory

Transport and server settings:

Var	Default	Meaning
`HOST`	`0.0.0.0`	Bind address (HTTP transports)
`PORT`	`8000`	Bind port (HTTP transports)
`TRANSPORT`	`stdio`	`stdio`, `streamable-http`, or `sse`
`ENABLE_OTEL`	`True`	OpenTelemetry / Langfuse export
`EUNOMIA_TYPE`	`none`	Access-governance mode: `none`, `embedded`, `remote`

Every per-resource tool can be toggled with its *TOOL switch (for example INVENTORYTOOL, JOBSTOOL, JOB_TEMPLATESTOOL, SYSTEMTOOL). The full set, with defaults, is documented in .env.example. Copy it to .env and fill in only what you use.

Backing Service¶

The Ansible Tower / AWX controller this connector targets is an external Ansible Automation Platform — Red Hat Ansible Automation Platform is a managed / commercial product, and the upstream AWX project is deployed through the AWX Operator on Kubernetes. This package does not provision the controller; only connection configuration (the ANSIBLE_* variables above) is required. Point ANSIBLE_BASE_URL at an already-running controller and supply credentials.

Docker Compose¶

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

services:
  ansible-tower-mcp-mcp:
    image: example/ansible-tower-mcp@sha256:<digest>
    container_name: ansible-tower-mcp-mcp
    hostname: ansible-tower-mcp-mcp
    restart: always
    env_file:
      - ../.env
    environment:
      - PYTHONUNBUFFERED=1
      - HOST=0.0.0.0
      - PORT=8000
      - TRANSPORT=streamable-http
    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 set ANSIBLE_* values
docker compose -f docker/mcp.compose.yml up -d
docker compose -f docker/mcp.compose.yml logs -f

Run the agent server¶

The companion A2A agent server is the ansible-tower-agent console script. It is a graph-routed Pydantic-AI agent that connects to the MCP server over MCP_URL and exposes an AG-UI web interface and an A2A endpoint on its own port (default 9012):

export MCP_URL=http://localhost:8000/mcp
export PROVIDER=openai
export MODEL_ID=gpt-4o
ansible-tower-agent --host 0.0.0.0 --port 9012

The repo ships docker/agent.compose.yml, which runs the MCP server and the agent server together. The agent service wires MCP_URL to the MCP container and publishes the agent on :9012:

services:
  ansible-tower-mcp-agent:
    image: example/ansible-tower-mcp@sha256:<digest>
    container_name: ansible-tower-mcp-agent
    depends_on:
      - ansible-tower-mcp-mcp
    command: [ "ansible-tower-agent" ]
    environment:
      - HOST=0.0.0.0
      - PORT=9012
      - MCP_URL=http://ansible-tower-mcp-mcp:8000/mcp
      - PROVIDER=${PROVIDER:-openai}
      - MODEL_ID=${MODEL_ID:-gpt-4o}
      - ENABLE_WEB_UI=True
    ports:
      - "9012:9012"

docker compose -f docker/agent.compose.yml up -d

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
ansible-tower-mcp.example.invalid {
    tls internal
    reverse_proxy ansible-tower-mcp-mcp:8000
}

# Public — automatic Let's Encrypt
ansible-tower-mcp.example.com {
    reverse_proxy ansible-tower-mcp-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=ansible-tower-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 ansible-tower-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:

{
  "mcpServers": {
    "ansible-tower-mcp": {
      "command": "uv",
      "args": ["run", "ansible-tower-mcp"],
      "env": {
        "ANSIBLE_BASE_URL": "https://your-tower.example.com",
        "ANSIBLE_USERNAME": "admin",
        "ANSIBLE_PASSWORD": "secret",
        "ANSIBLE_TOWER_TLS_PROFILE": "private-pki"
      }
    }
  }
}

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