Deployment¶

Deployment Options¶

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

Loopback development listener¶

camunda-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/camunda-mcp@sha256:<digest> camunda-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": {
    "camunda": {"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 camunda-mcp as long-lived servers: the transports, the and giving it a DNS name with Technitium. To provision the Camunda platform it connects to, see Backing Platform.

camunda-mcp ships both an MCP server (console script camunda-mcp) and an deterministic tool surface a policy router or agent calls; the agent server wraps that surface in a Pydantic-AI graph agent for agent-to-agent orchestration.

Run the MCP server¶

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

stdio (default)streamable-httpsse

camunda-mcp

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

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

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

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

Health check (HTTP transports):

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

Configuration (environment)¶

camunda-mcp is configured entirely from the environment. The required set depends on which platform the operator target (CAMUNDA_PLATFORM):

Var	Default	Meaning
`CAMUNDA_PLATFORM`	`7`	Target platform: `7` or `8`
`CAMUNDA_TLS_PROFILE`	`system`	Named outbound TLS policy from AgentConfig
`CAMUNDA_TIMEOUT_SECONDS`	`30`	Finite connect/read timeout for all Camunda HTTP calls
`CAMUNDA_MAX_RESPONSE_BYTES`	`8388608`	Maximum accepted response body size
`CAMUNDA7_URL`	`http://localhost:8080/engine-rest`	Camunda 7 Engine REST base URL
`CAMUNDA7_TOKEN`	—	Camunda 7 bearer token (optional)
`CAMUNDA7_USERNAME`	—	Camunda 7 basic-auth user (optional)
`CAMUNDA7_PASSWORD`	—	Camunda 7 basic-auth password (optional)
`CAMUNDA8_ZEEBE_REST_URL`	`http://localhost:8080`	Camunda 8 Zeebe REST base URL
`CAMUNDA8_OPERATE_URL`	—	Camunda 8 Operate REST URL
`CAMUNDA8_TASKLIST_URL`	—	Camunda 8 Tasklist REST URL
`CAMUNDA8_CLIENT_ID`	—	Camunda 8 OAuth client id
`CAMUNDA8_CLIENT_SECRET`	—	Camunda 8 OAuth client secret
`CAMUNDA8_OAUTH_URL`	—	Camunda 8 OAuth token URL
`CAMUNDA8_AUDIENCE`	`zeebe.camunda.io`	Camunda 8 OAuth audience
`CAMUNDATOOL`	`True`	Register the Camunda tool set

Plus HOST / PORT / TRANSPORT for HTTP transports. Credentials left blank leave the corresponding platform inactive — the server remains inactive when credentials are absent. A starter .env.example ships with the repository; copy it to .env and fill in the values for the platform the operator use.

Docker Compose¶

The repository ships docker/mcp.compose.yml. A representative stack reads a sibling .env and publishes the HTTP server on :8000:

services:
  camunda-mcp:
    image: example/camunda-mcp@sha256:<digest>
    container_name: camunda-mcp
    hostname: camunda-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 edit CAMUNDA_* values
docker compose -f docker/mcp.compose.yml up -d
docker compose -f docker/mcp.compose.yml logs -f

Run the A2A agent server¶

camunda-mcp also ships an agent server (console script camunda-agent, described by a2a.json). It wraps the MCP tool surface in a Pydantic-AI graph agent and is wired to a running MCP server through the MCP_URL environment variable:

export MCP_URL=http://camunda-mcp:8000/mcp
export DEFAULT_AGENT_NAME="Camunda Agent"
camunda-agent --host 0.0.0.0 --port 8001

The repository ships A representative agent stack publishes the agent on :8001 and points it at the MCP server:

services:
    image: example/camunda-mcp@sha256:<digest>
    container_name: camunda-agent
    hostname: camunda-agent
    restart: always
    command: ["camunda-agent", "--host", "0.0.0.0", "--port", "8001"]
    env_file:
      - .env
    environment:
      - PYTHONUNBUFFERED=1
      - MCP_URL=http://camunda-mcp:8000/mcp
    ports:
      - "8001:8001"
    depends_on:
      - camunda-mcp

Behind a Caddy reverse proxy¶

Expose the HTTP server on a hostname with automatic TLS. Add to the operator's Caddyfile:

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

# Public — automatic Let's Encrypt
camunda-mcp.example.com {
    reverse_proxy camunda-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=camunda-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 camunda-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 the operator's client's mcp_config.json (multiplexer nickname camun):

{
  "mcpServers": {
    "camunda-mcp": {
      "command": "uv",
      "args": ["run", "camunda-mcp"],
      "env": {
        "CAMUNDA_PLATFORM": "7",
        "CAMUNDA7_URL": "http://your-camunda:8080/engine-rest",
        "CAMUNDATOOL": "True"
      }
    }
  }
}

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