Executors

An executor performs the actual work for a run. Wakeplane ships three executor types. The target type is declared on the schedule; the dispatcher routes to the correct executor at dispatch time.

HTTP executor

Makes an HTTP request to a URL.

{
  "target_type": "http",
  "target": {
    "url": "https://internal.example.com/hooks/daily-report",
    "method": "POST",
    "headers": {
      "Content-Type": "application/json"
    },
    "body": "{\"triggered_by\": \"wakeplane\"}"
  }
}

The executor records the response status, body (truncated), and latency in the run receipt.

Shell executor

Runs a shell command.

{
  "target_type": "shell",
  "target": {
    "command": "/usr/local/bin/run-report.sh",
    "args": ["--env", "production"],
    "timeout": "5m"
  }
}

Stdout/stderr are captured in the run receipt. The exit code determines run state.

Workflow executor

Calls an in-process Go function registered in the workflow registry.

app.RegisterWorkflow("generate-report", func(ctx context.Context, payload []byte) error {
    return generateReport(ctx, payload)
})

{
  "target_type": "workflow",
  "target": {
    "name": "generate-report",
    "payload": "{}"
  }
}

Workflow handlers must be registered at startup. Dynamic or out-of-process loading is not supported in v0.1.

Executor contract

All executors:

Run only after a durable claim exists
Respect the run context for cancellation
Record a result and receipt before the dispatcher transitions run state
Do not modify schedule or occurrence state directly

Current limits

Workflow handlers that ignore ctx.Done() are non-cooperative. The dispatcher enforces a timeout boundary and reports remaining active workers on shutdown. The caller decides whether to wait or let process supervision handle termination.