Camunda Orchestration Cluster MCP Server – Workflow Ops
Camunda Orchestration Cluster exposes an MCP server built into Camunda 8.9 and later so AI agents can discover and invoke operational workflow tools. Agents can inspect cluster health, work with incidents, process definitions, process instances, user tasks, and variables using the same authentication and authorization model as the Orchestration Cluster API.
Overview
The Camunda Orchestration Cluster MCP Server is a built-in Model Context Protocol API surface for Camunda 8 Orchestration Cluster. It lets MCP-capable AI agents and LLM applications access Camunda operational capabilities through tool discovery instead of custom REST integration code. The server is documented for Camunda 8.9 and later and is intended for operational workflows around running Camunda clusters, processes, incidents, tasks, and variables.
What the MCP server enables
The server exposes discoverable MCP tools across several Camunda domains. Agents can check cluster health and topology, search and retrieve incidents, resolve incidents, search process definitions, retrieve BPMN XML, search and create process instances, search and assign user tasks, inspect user-task variables, and search or retrieve variables. Tool names, parameters, and response schemas are discoverable at runtime by MCP clients, and Camunda notes that exact tool signatures may evolve across versions.
When to use it
Use this MCP server when an AI assistant needs live operational context from Camunda rather than static documentation. Practical examples include investigating failed workflow executions, retrieving incident details, checking partition health, finding a deployed process definition, starting a process instance with variables, or assigning a pending user task during a troubleshooting session. It is also useful inside Camunda agentic orchestration workflows through Camunda MCP Client connectors, where an AI agent running in a BPMN process can query incidents or start processes as part of automation.
Connection and authentication
The native remote transport is Streamable HTTP at the /mcp/cluster endpoint. Local Camunda 8 Run and Docker Compose deployments expose http://localhost:8080/mcp/cluster and are enabled by default. SaaS clusters use a public endpoint of https://${REGION_ID}.api.camunda.io/${CLUSTER_ID}/mcp/cluster, or a private connectivity endpoint ending in /mcp/cluster. Authenticated environments use the same model as the Orchestration Cluster REST API: OAuth 2.0 client credentials with the Orchestration Cluster API scope, sent as Authorization: Bearer <token>. For MCP clients that cannot perform OAuth client credentials directly, Camunda documents the c8ctl mcp-proxy STDIO bridge using npx -y @camunda8/cli mcp-proxy.
Key considerations
The Orchestration Cluster MCP Server is not part of the Camunda 8 public API. It is available from Camunda 8.9 onward, enabled by default only for Camunda 8 Run and Docker Compose, and must be enabled explicitly for production deployment types such as SaaS or Helm/Kubernetes. Permissions are governed by Camunda authentication and authorization, so API clients should be scoped carefully and secrets should be stored in the MCP host or Camunda secret mechanism rather than embedded in configuration files.
Supported Transports
streamable_http
URL: https://${REGION_ID}.api.camunda.io/${CLUSTER_ID}/mcp/cluster
stdio
Command: npx
Args:
-y@camunda8/climcp-proxy
Frequently Asked Questions
- When should an AI agent use the Camunda Orchestration Cluster MCP Server?
- Use it when an agent needs to inspect or act on live Camunda operational data, such as checking cluster health, investigating incidents, retrieving BPMN XML, creating process instances, assigning user tasks, or reading variables during workflow troubleshooting and orchestration.
- What does this MCP server add to an AI agent's capabilities?
- It gives the agent runtime-discoverable Camunda tools over MCP, so the agent can work with cluster status, topology, incidents, process definitions, process instances, user tasks, and variables without implementing bespoke REST API calls.
- What can the AI agent access or manage through this server?
- Documented tools cover cluster health and topology, incident search/retrieval/resolution, process definition search and BPMN XML retrieval, process instance search/retrieval/creation, user task search/retrieval/assignment, user-task variables, and variable search or retrieval.
- How is authentication configured for this MCP server?
- The remote HTTP server uses the same authentication model as the Camunda Orchestration Cluster REST API. SaaS users create API client credentials with the Orchestration Cluster API scope, obtain an OAuth 2.0 access token, and send it with an Authorization bearer header; the documented c8ctl proxy can handle this flow for clients that only support STDIO.
- Which transport should be used for this MCP server?
- Use Streamable HTTP when the MCP client supports remote HTTP and can provide the required authentication. Use the documented STDIO proxy through npx and @camunda8/cli mcp-proxy when the MCP client does not natively support OAuth 2.0 client credentials for the remote Camunda endpoint.