# Start

Node type `start`

## What it does

Read initial parameters and declare custom variables before the first action.

## Desktop client configuration

![Initial parameters in the desktop client](/assets/start-en.7f48eae7a9dbcfe5258cb5ef5dfa68d1f735d164365a42959befc39f083bc2f6.0538f3a9.webp)

Captured in the English interface of MoreLogin Desktop V2.73.101.

## Inputs

Paths are relative to the complete node.

| Field | Type or values | Schema required | Execution default | Description and bounds |
|  --- | --- | --- | --- | --- |
| `id` | `string` | Yes | — | `minLength = 1`, `maxLength = 128` |
| `meta` | `object` | No | — | Optional opaque display JSON. Unknown nested keys are preserved; no runtime semantics. Omission on replacement removes it; untouched nodes retain it. |
| `type` | `"start"` | Yes | — | — |
| `data` | `object` | No | — | — |
| `data.title` | `string` | No | — | — |
| `data.description` | `string` | No | — | — |
| `data.inputs` | `object` | No | — | — |
| `data.outputs` | `object` | No | — | — |
| `data.customVariables` | array<`object`> | No | — | — |
| `data.inputsValues` | `object` | No | — | — |
| `data.taskSettings` | `object` | No | — | Template task settings. They take effect only in full runs (a whole-workflow run in the editor, a run started through the Local API, or a scheduled task). They have no effect in single-step or group debugging, and the settings in a called child template have no effect. When omitted, all settings use their defaults. |
| `data.taskSettings.completionTabs` | `"keep"` / `"closeOthers"` / `"clear"` | No | — | How the main environment's tabs are handled after the task completes: keep=keep the tabs; closeOthers=keep only the current tab; clear=clear the tabs (closes the browser directly). Default: keep. |
| `data.taskSettings.completionBrowser` | `"keep"` / `"close"` | No | — | What happens to the main environment's browser after the task completes: keep=keep it open; close=close it. Nothing is done when the user stops the run manually. Default: keep. |
| `data.taskSettings.headless` | `boolean` | No | — | Starts the browsers that this run opens in headless mode (requires browser kernel 130 or later). They are always closed when the task ends. Environments that were already open before the run are not affected. Default: false. |
| `data.taskSettings.disableAutofill` | `boolean` | No | — | Browsers that this run opens do not get the environment's account passwords injected and do not autofill. Default: false. |
| `data.taskSettings.allowSavePasswords` | `boolean` | No | — | When false, account changes in browsers that this run opens are not saved to the environment. Default: false. |
| `data.taskSettings.clearCacheOnClose` | `boolean` | No | — | After a browser that this run opened closes (no matter who closes it), all local cache of that environment is cleared. The cloud cache is not cleared. Default: false. |
| `data.taskSettings.timeoutSeconds` | `integer` | No | — | Task timeout in seconds; 0 means no limit. The timer starts when the delay ends and the browser starts. On timeout, the main environment's browser is closed and the run fails with RPA_TASK_TIMEOUT. Default: 0. `minimum = 0` |
| `data.taskSettings.delayMinutes` | `integer` | No | — | Delayed start in minutes; 0 means no delay. During the delay, the run stays queued and does not occupy the environment; see the run's startAfter. Default: 0. `minimum = 0` |
| `data.taskSettings.launchArguments` | `string` | No | — | Browser launch arguments, one per line. An argument with the same name takes precedence over the environment's own argument. Blocklisted arguments, and arguments such as the debugging port, extension loading, user data directory, and headless switch, have no effect. Default: empty. |
| `blocks` | `array` | No | — | `maxItems = 0` |


## Mode requirements and limits

- This node must be the first root node, and only one is allowed. It declares customVariables and provides no business output.
- Field paths of structure nodes are relative to the complete node. For end, defaults refers to data.inputsValues. For other details, see workflow.md.
- Validation code `RPA_STRUCTURE_INVALID`


## Outputs

No dedicated business output.

## JSON example

Place the `start` node first in the root `nodes` array.

```json
{
  "id": "start",
  "type": "start",
  "data": {
    "title": "Set up the attempt counter",
    "customVariables": [
      {
        "id": "attempts",
        "name": "Attempts",
        "type": "number",
        "initialValue": 0
      }
    ]
  }
}
```

[Back to the node catalog](/browser-rpa/nodes)