See llms.txt for all machine-readable content.

Back to Templates

Moderate toxic text, redact PII, and audit with a Gemini HTTP API

Created by

Created by: Oka Hironobu || okp29
Oka Hironobu

Last update

Last update 17 hours ago

Categories

Share

Quick Overview

This workflow exposes a POST webhook that sends user text (and optional context) to Google Gemini for moderation, returning an allow/flag/block decision, toxicity level, matched categories, detected PII types, a short reason, and a cleaned version with slurs masked and PII redacted.

How it works

  1. Receives a POST request on a webhook endpoint with a JSON body containing text and optional context.
  2. Sends the input to Google Gemini with instructions to classify safety (allow/flag/block), rate toxicity, identify categories, detect PII, and produce a redacted/masked cleaned_text.
  3. Parses Gemini’s response into a structured JSON object with the required moderation fields.
  4. Returns the moderation result to the caller as the webhook response.
  5. If Gemini fails to classify the text, returns a fail-safe JSON response that flags the content for human review.

Setup

  1. Add a Google Gemini (PaLM) API credential for the Google Gemini Chat Model node.
  2. Activate the workflow and copy the production webhook URL from the webhook trigger.
  3. Configure your client/app to POST JSON (for example { "text": "...", "context": "..." }) to the /moderate endpoint and handle the JSON response.