Manage Workflows with the JumpCloud MCP Server for Admins
Use the JumpCloud Model Context Protocol (MCP) Server for admins to create, inspect, modify, and troubleshoot Workflows from a connected AI client. Describe the automation you need in natural language, review the workflow definition, and inspect execution results in the same conversation.
Workflows support is available in Public Preview through the existing JumpCloud MCP Server. If your client is already connected, you do not need a separate Workflows connection or additional credentials. This article covers using an external MCP client.
Workflows created through MCP are also editable in the Workflow Builder. You can use your AI client and the Admin Portal to inspect and manage the same workflow.
Prerequisites
- Enable the MCP Server for admins and connect a supported AI client. See Get Started: JumpCloud MCP Server for Admins for setup and authentication instructions.
- Use an admin account with permission to manage Workflows. The MCP tools inherit the existing server's permission scoping.
- Create any groups, policies, commands, or notification channels required by your workflow. See Get Started: Workflows.
Discovering and Inspecting Workflows
Ask your AI client to list the workflows in your organization. Select a workflow, then ask for its full definition to review its trigger configuration, conditions, action nodes, and variable mappings.
The tools for discovery and inspection are:
workflows_list: List workflows in your organization.workflow_get: Retrieve a specific workflow's definition.
For example:
List the workflows in my JumpCloud organization and show their status. Retrieve the definition of the workflow I select and explain its trigger, conditions, actions, and variable mappings.
Use the workflow ID returned by the server to identify the workflow in follow-up requests.
Creating and Modifying Workflows
Use workflow_create to create a workflow from a structured definition. Use workflow_update to modify an existing workflow. Your AI client can help translate your request into the Workflows domain-specific language (DSL).
When describing an automation, include:
- The event, schedule, or manual trigger that starts it.
- The conditions that determine when actions run.
- The actions to perform and the resources they affect.
- The values each action needs from the trigger or earlier actions.
For example:
Help me draft a workflow that adds a newly created user to a specific group when the user's attributes match my criteria. Ask me for the group and criteria. Show me the proposed trigger, conditions, actions, and DSL before creating it.
For an existing workflow:
Retrieve workflow WORKFLOW_ID and explain its current behavior. Propose the changes needed to update its conditions. Show me the revised definition before applying changes.
Replace WORKFLOW_ID with the ID returned by the server. Review the proposed definition and resolve missing configuration before requesting a create or update operation.
When the draft is ready, explicitly ask your client to create the workflow. Proposing a definition does not create a saved workflow in your organization.
Newly created workflows start in an inactive state so you can review their configuration before enabling them.
Inspecting the Created Workflow in the Admin Portal
Open the created workflow in the Admin Portal. On the Overview tab, inspect its trigger, conditions, action cards, and variable mappings. Select the trigger to view its event details and filter conditions in the Trigger Info panel.
Check the workflow's current status before enabling it. You can also ask your AI client to retrieve the saved definition and explain its behavior. For portal management instructions, see Manage Workflows.
For example:
Retrieve workflow WORKFLOW_ID and explain when it runs, which conditions it checks, and what each action changes. Show its current status and execution role.
Duplicating a Workflow
Use workflow_duplicate to copy an existing workflow as the starting point for a variation. The duplicate is created as a disabled copy. Duplication is available through MCP even though it is not currently available in the Workflows UI.
For example:
Retrieve workflow WORKFLOW_ID and summarize its configuration. Help me duplicate it for another group. Ask me which group to use and show the proposed changes before duplicating or updating anything.
Review the duplicate's definition and status before enabling it. Ask your client to update any configuration needed for the new use case.
Validating and Testing a Workflow
The Workflows API validates DSL when you create or update a workflow. If a write returns a validation error, ask your client to explain it and propose a corrected definition for review.
For example:
Explain the validation error returned when creating this workflow. Show the affected configuration and propose a corrected definition. Wait for my approval before submitting the change.
Preparing Test Resources
Ask your AI client to help prepare resources for a test scenario. For example, it can create a dedicated test user and identify existing test devices in your organization. Review the proposed resources and actions before approving setup or execution.
Important:
Testing can make live changes in your organization. Workflow actions can modify users, change device bindings, or delete device records. Review which resources each action affects before approving the test.
For example:
Help me plan a test for workflow WORKFLOW_ID. Identify the test resources needed and the actions the workflow will perform. Use a dedicated test user and existing test devices. Show me the setup and execution plan before making changes.
Running and Inspecting a Test
Ask your client to run the approved test scenario. For an event-triggered workflow, the test must produce the configured event. For a workflow with an external-source trigger, use workflow_trigger_run with the required input to manually initiate a run.
For example:
Retrieve workflow WORKFLOW_ID and confirm whether its trigger permits a manual run. Show me the required input and actions it will perform. Wait for my confirmation before starting the run, then retrieve the result.
Use workflow_runs_list to find the execution and workflow_run_get to inspect its results. Review the run status, per-node outcomes, and error details to check whether the workflow behaved as intended.
You can also open the workflow in the Admin Portal and select Execution history. Select a run to inspect its Execution results, including completed actions and actions skipped because their conditions did not match.
Managing the Workflow Lifecycle
Use workflow_update to enable or disable a workflow through a status update. Ask your client to retrieve the definition and explain its actions before enabling it.
For example:
Retrieve workflow WORKFLOW_ID and show its status. Explain the trigger and actions, then ask me to confirm before enabling it.
To stop automatic execution, ask your client to disable the workflow and retrieve its updated status.
Use workflow_delete to delete a workflow. Identify the target workflow and review its definition before requesting deletion.
For example:
Retrieve workflow WORKFLOW_ID and show its name and configuration. Ask me to confirm the target before deleting it.
Troubleshooting Workflow Runs
Use the run tools to investigate an execution without leaving the conversation where you built the workflow:
workflow_runs_list: Query workflow execution history.workflow_run_get: Retrieve a specific run's status, per-node outcomes, and error detail.
For example:
Show the recent runs for workflow WORKFLOW_ID. Retrieve a failed run, identify the failed node, and explain the returned error. Compare the failure with the workflow definition and propose a correction for me to review.
Use the returned run detail to guide the next request. Ask the client to inspect the relevant trigger configuration, expression, variable mapping, or action parameters. Review the proposed change before updating the workflow, then inspect a subsequent run to verify the result.
Was this information helpful?