StealthGPT icon

StealthGPT

Generate natural, human, publish-ready long-form content or humanize existing text with StealthGPT

Actions4

Text Humanization → Start Run

AI-generated

Summary

Starts a StealthGPT text humanization run that rewrites supplied text into natural, human-sounding writing. By default it polls until the run finishes and returns the final result. It can instead return the run ID right away.

Inputs

  • Text (required) — Text to humanize. The run fails with an error if it is empty or whitespace-only.
  • Quality Mode — Quality (default) or Fast. Trades output quality against speed.
  • Model — Heavy (default) or Lite humanization model.
  • Output Format — Text (default) or Markdown format for the returned result.
  • Wait for Completion — Default on. When on, polls the run until it reaches a terminal status. When off, returns the creation response (with the run ID) immediately.
  • Poll Interval (Seconds) — Seconds between status checks while waiting. Default 5, minimum 1. Shown only when Wait for Completion is on.
  • Max Wait (Seconds) — Maximum time to wait. Default 600, minimum 1. The node errors on timeout, but the run may still complete. Shown only when Wait for Completion is on.
  • Options — Optional collection. Idempotency Key is sent as an idempotency-key header so retries don't create duplicate runs or charges. Webhook URL makes StealthGPT call that URL with the final result or failure. Webhook Secret signs the webhook payload. Empty values are omitted.

Output shape

One JSON item per input item. With Wait for Completion on, it is the final run status object from GET /api/stealthify/runs/{runId} with status 'completed'. With it off, it is the response from the create call (POST /api/stealthify/runs), which includes the runId.

If the run ends as failed or cancelled, the node throws an error using the API's error message and code. With Continue On Fail enabled, it outputs {error: message} for that item instead. A wait timeout also throws, and the message includes the run ID so you can fetch the result later with Get Run Status. Input items are processed one at a time. The exact result fields come from the StealthGPT API and are not defined in the node source.

Examples

Example 1: Humanize AI-drafted text and use the result immediately in the next node.

Text set to an expression such as {{$json.draft}}, Quality Mode Quality, Model Heavy, Wait for Completion on, with the default poll interval and max wait.

Example 2: Start a long run without blocking, then check it later.

Wait for Completion off. Store the returned runId and fetch the result later with Text Humanization > Get Run Status.

Example 3: Send the result to another system by webhook, with safe retries.

Wait for Completion off. In Options, set Webhook URL, a Webhook Secret of at least 16 characters, and an Idempotency Key.

Links

Discussion