Zammad MCP Server — Deployment
Docker, HTTP transport, and production deployment options for the Zammad MCP Server.
Deployment
This page covers common deployment patterns. For cloud-specific guides (Cloud Run, Fly.io, Railway), see the full Deployment guide on GitHub.
Local development with Zammad
The repository includes Docker Compose for a local Zammad instance. Default:
Zammad 7.2.2. Set ZAMMAD_VERSION in docker/.env for 7.1, 7.0, 6.5, or 6.3
(see COMPATIBILITY.md):
git clone https://github.com/openticketai/zammad-mcp-server.gitcd zammad-mcp-server/dockerdocker-compose up -dWait until Zammad is ready (often 2–3 minutes), then open http://localhost:8080, create an API token,
and point ZAMMAD_URL at that host from your MCP client.
Install from source
git clone https://github.com/openticketai/zammad-mcp-server.gitcd zammad-mcp-serveruv sync --extra devzammad-mcp-serverDocker image
Build and run the packaged server:
docker build -t zammad-mcp-server .docker run --rm -e ZAMMAD_URL=... -e ZAMMAD_HTTP_TOKEN=... zammad-mcp-serverPass environment variables for URL, token, and MCP_DENIED_TOOLS. Mount no ticket data volumes — the
server is stateless.
HTTP transport (remote MCP)
zammad-mcp-server --transport http --host 0.0.0.0 --port 8000The Docker image defaults to this transport. MCP path is /mcp/. There is no /health HTTP route — check that port 8000 is listening.
SSE transport (legacy)
zammad-mcp-server --transport sse --host 0.0.0.0 --port 8000Use only for older MCP clients. Restrict network access; terminate TLS at a reverse proxy in production.
PyPI releases
Production installs typically use pinned versions:
pip install zammad-mcp-server==0.2.0# oruv tool install zammad-mcp-server==0.2.0Check PyPI for the current release. Pre-built
container images are published on GitHub Container Registry:
ghcr.io/openticketai/zammad-mcp-server.
CI and quality
The public repository runs automated tests and lint on each push. Before relying on a release in production, verify the GitHub Actions status on the tag you deploy.
Architecture
For component diagrams, request flow, and extension points, read ARCHITECTURE.md on GitHub.
