LoopQuest icon

LoopQuest

Submit AI output for human-in-the-loop review

Actions2

Create Review Task

AI-generated

Summary

Create a LoopQuest review task for a selected review game, optionally pausing the workflow for a human verdict or reviewing in the background. The operation sends a POST request for each input item and returns the API response for each task.

Inputs

  • Module (required) — Select the review game: Swiper approves or flags content; Detective finds errors in text; Redact masks sensitive data; Grounding checks a claim against source text; Sorter classifies into buckets; Fixer confirms or corrects fields; Versus compares two answers.
  • Content — AI output used by Swiper, Detective, and Redact. For Grounding, it serves as the claim when Claim is empty. The source builds default payloads for Fixer and Versus too, but their payloads generally need Payload JSON.
  • Claim — For Grounding, the claim to verify. If empty, Content is used instead.
  • Source Text — For Grounding, the source or context against which the claim is checked.
  • Buckets — For Sorter, comma-separated categories; whitespace and empty entries are removed before sending.
  • Mode — Gate waits for a verdict; Monitor submits a background review without blocking. Defaults to Monitor.
  • Additional Fields — Optional task settings: Title for the review card; Source label (defaults to n8n); Callback URL for signed verdict webhooks; External ID echoed in the verdict webhook; Idempotency Key sent as a request header to deduplicate retries; Reviewers Required (defaults to 1); and, in Gate mode, Timeout (Seconds) and On Timeout (defaults to Escalate). Payload JSON supplies a raw game-specific payload and overrides the fields constructed from the other inputs; the UI description gives Versus and Fixer examples.

Output shape

A list of task API response items, one per input item.

Each response is returned as an n8n item's json data and paired with its input item. The exact response fields depend on the LoopQuest API.

Examples

Example 1: Swiper review

Select Swiper, provide Content, and choose Gate or Monitor.

Example 2: Grounding review

Select Grounding and provide Source Text plus Claim, or leave Claim empty to use Content.

Example 3: Sorter review

Select Sorter and provide comma-separated Buckets.

Links

Discussion