Guide

Content Moderation

Filter toxic reviews automatically with Perspective API or OpenAI.

nuxt-reviews includes built-in content moderation to automatically filter or flag toxic reviews.

Setup

nuxt.config.ts

reviews: {
  moderation: {
    enabled: true,
    provider: 'perspective',
    apiKey: process.env.PERSPECTIVE_API_KEY,
    toxicityThreshold: 0.7,
    action: 'filter'
  }
}

Options

Option	Type	Default	Description
`enabled`	`boolean`	`false`	Enable content moderation
`provider`	`'perspective' \| 'openai'`	—	Moderation provider
`apiKey`	`string`	—	Provider API key
`toxicityThreshold`	`number`	`0.7`	Score threshold (0.0 - 1.0)
`attributes`	`PerspectiveAttribute[]`	All	Perspective API only
`action`	`'filter' \| 'flag'`	`'filter'`	Action on flagged reviews

Filter vs Flag

Filter Mode

Toxic reviews are removed from the response entirely:

moderation: {
  action: 'filter' // Toxic reviews are excluded
}

Flag Mode

Reviews are kept but include moderation metadata:

moderation: {
  action: 'flag' // Reviews include moderation data
}

In flag mode, each review includes a moderation field:

{
  "moderation": {
    "flagged": true,
    "provider": "perspective",
    "scores": {
      "toxicity": 0.85,
      "severeToxicity": 0.12,
      "insult": 0.78,
      "profanity": 0.65,
      "threat": 0.02,
      "identityAttack": 0.05
    }
  }
}

Perspective API Attributes

moderation: {
  attributes: ['TOXICITY', 'SEVERE_TOXICITY', 'INSULT', 'PROFANITY']
}

Available: TOXICITY, SEVERE_TOXICITY, INSULT, PROFANITY, THREAT, IDENTITY_ATTACK.

Composables

useReviews() and useReviewSchema() — reactive, type-safe review data and Schema.org JSON-LD.

Component Kit

Four ready-to-use Vue components — ReviewStars, ReviewCard, ReviewList, ReviewSummary.