Camunda (BPMN Engine) MCP. Orchestrate and debug complex business workflows.

This server lets your AI client act like a technical process orchestrator for Camunda. You can use it to manage the whole lifecycle of a business workflow, from defining the process to handling failures.

Start a New Process

Kick off a new workflow instance using start_process_instance and pass any variables it needs. You can also use search_process_instances to find existing running processes, and search_process_definitions to find the models themselves.

Handle User Tasks

If a human has to make a decision, you can search for pending tasks using search_user_tasks, then grab the details of a specific task with get_user_task, or pull the necessary form using get_user_task_form. You can assign a task to a user with assign_user_task, and once they finish, you mark the task complete using complete_user_task and send the data needed to move the process forward.

You can also remove a task assignment using unassign_user_task.

Manage Jobs and Failures

For background jobs, you can locate failures by searching with search_jobs, then activating jobs with activate_jobs or completing them with complete_job. If things go sideways, you can manually mark a job as failed using fail_job or simulate an error with throw_job_error. You can also check the status of a job or instance using get_process_instance or search_jobs.

Inspect System State and Logic

To understand the environment, you can check the cluster health with get_topology or search for users and groups using search_users and search_groups. You can view the raw workflow logic by fetching the BPMN XML definition for a process using get_process_definition_xml, and you can upload new models using deploy_resources.

Troubleshoot Incidents

If a process breaks, you can search for incidents using search_incidents, and then get the specific details of a problem using get_incident. You can also check the status of variables across the system using get_variable or searching with search_variables.