Rundeck

Run and control Rundeck jobs, adhoc commands, and scripts, and inspect executions from your flows over the Rundeck API. Runbook automation and job scheduling for operations.

15 actions available

An upstream flow signals a runbook should run

Agent resolves the project and the job to run, plus its option values

Agent confirms the job and the target environment match what the flow expects

Agent posts the job, its options, and its target to the operations channel

An operator approves the run when the runbook is destructive or targets production

On approval, the agent calls Run Job and polls Get Execution State to completion

What This Integration Enables

FlowRunner agents use Rundeck to run operational runbooks from inside a larger flow. An agent can trigger a job with option values and track the resulting execution to completion, kick off deployment or maintenance runbooks and poll their status, state, and log output, and run adhoc shell commands or scripts across project nodes without defining a job. It can retry failed executions, abort runaway runs, and audit or clean up execution history. Run Job returns the execution immediately, so the agent polls its state to follow it. What it does not do on its own is run a destructive runbook against production. That approval stays with the operator who owns the environment.

Without FlowRunner

Manual runbook runs An operator opens Rundeck and runs each job by hand

Options by memory Job option values are re-typed each run and easy to get wrong

Unlogged production runs A destructive runbook fires against production with no recorded approval

With FlowRunner

Triggered from flows The agent runs the runbook as part of the larger workflow

Options passed in The agent supplies the option values the flow already knows

Approved production runs Destructive or production runbooks wait for an operator to approve

Use Case Scenarios

Deploy runbook after a build

A build completes upstream. The agent calls Run Job to execute the deployment runbook for the new artifact, passing the version as an option value, then polls Get Execution State until it finishes. On success it posts the result; on failure it captures the logs with Get Execution Output. The runbook runs as part of the pipeline instead of as a separate manual step.

Adhoc diagnostics across nodes

An incident flow needs a quick read across the fleet. The agent calls Run Adhoc Command with a node filter to gather the same diagnostic from each matching node, then summarizes the output. Because the command is read-only, the agent runs it on its own and hands the responder a consolidated view.

Destructive runbook with an approval gate

A cleanup flow needs a runbook that drops data or restarts a production service. The agent resolves the job, its options, and its target, then posts them to the operations channel. It does not run it. Only after an operator approves does the agent call Run Job, and it records the approval and the execution ID in the flow log.

Human-in-Loop Highlight

The human-in-loop moment in Rundeck is the destructive-run gate. Listing projects and jobs, running read-only adhoc commands, and polling executions are all safe to automate, so the agent runs them by rule. Running a destructive runbook against production is not, because a runbook that drops data or restarts a service cannot be walked back once it starts. When a flow reaches a job flagged destructive or targeting production, the agent posts the job, its option values, and its target and asks in the operations channel: "Ready to run the purge-old-tenants job against prod with retention=30d. Approve?" An operator approves. Only then does the agent call Run Job and poll it to completion. The agent runs safe runbooks on its own. A person owns the run that changes production.

Agent processes routinely

Detects exception requiring judgment

Clear match Continues automatically

Ambiguous Routes to human via preferred channel

Human decides

Agent resumes with decision

Agent Capabilities

15 actions

Projects

List Projects List the Rundeck projects.
Get Project Retrieve a single project.

Jobs

List Jobs List the jobs in a project.
Get Job Definition Retrieve a job's full definition.
Run Job Start a job with option values. Returns the execution immediately without waiting for completion. Run destructive or production runbooks behind a human gate.
Retry Job Execution Re-run a job execution, optionally only its failed nodes.

Executions

Get Execution Retrieve an execution's details.
Get Execution State Retrieve an execution's node-by-node state. Used to poll a run to completion.
Get Execution Output Retrieve an execution's log output. Used to capture failure logs.
Abort Execution Abort a running execution.
List Project Executions List the recent executions in a project.
Delete Execution Delete an execution from history.

Adhoc

Run Adhoc Command Run a shell command across matching project nodes without a defined job.
Run Adhoc Script Run a script across matching project nodes without a defined job.

System

Get System Info Retrieve the server info. Used to confirm the URL, version, and token are valid.

Start building with Rundeck

$100 in credits. No card required. Connect in minutes.

Start Building Free Talk to us