Skip to main content
Runs represent active or historical executions of your workflows. Every time a workflow is triggered—whether by an event, a schedule, or a manual request—Thygon creates a new Run record. This record acts as the execution context, tracking the progress of each step, inputs, outputs, and any errors that occur.

Run lifecycle and states

The Runtime service orchestrates runs durably. A run transitions through the following statuses:
  • PENDING: The run has been queued and is waiting to be processed by the execution engine.
  • RUNNING: The runtime is actively executing steps in the workflow graph.
  • COMPLETED: All workflow steps finished successfully, and outputs have been generated.
  • FAILED: An error occurred during execution (e.g., node failure, timeout) that could not be recovered by retries.
               ┌───────────┐
               │  PENDING  │
               └───────────┘


               ┌───────────┐
               │  RUNNING  │
               └───────────┘

           ┌─────────┴─────────┐
           ▼                   ▼
     ┌───────────┐       ┌───────────┐
     │ COMPLETED │       │  FAILED   │
     └───────────┘       └───────────┘

Run variables and context

When triggering a run, you can pass custom parameters as input variables. The runtime makes these variables available to all nodes within the workflow:
{
  "runId": "run_01h3j4k5m6",
  "workflowId": "support_triage",
  "status": "RUNNING",
  "variables": {
    "priorityOverride": "high",
    "assigneeId": "usr_9988"
  }
}

Troubleshooting and observability

Because every step’s execution state, inputs, outputs, and error logs are linked to the run record, you can easily inspect and troubleshoot runs:
  • Step history: View the exact order and timestamp of executed nodes.
  • Node input/output: Inspect the data flowing between agents, deterministic steps, and connections.
  • Error details: Pinpoint which step failed and inspect the stack trace or error message returned.