Perspective API Shutdown: The Exact Code Changes You Need

TL;DR: Two things change when you swap from Perspective to Respectify’s compatible endpoint: the URL and the auth header. Everything else — your request body, attribute names, response parsing, thresholds — stays the same.

Google’s Perspective API is shutting down. If you have a running integration, you need to move. This post skips the theory and shows you the exact changes in TypeScript, Python, and PHP.

For the full migration story — what maps to what, how to handle spans, when to upgrade to full moderation — read the migration guide.

What changes (and what doesn’t)

Changes:

• Base URL: https://commentanalyzer.googleapis.com → https://app.respectify.org
• Path: /v1alpha1/comments:analyze → /v0.2/perspective-compat/analyse
• Auth: query-string ?key=YOUR_KEY → Authorization: Bearer YOUR_KEY header
• You’ll need a Respectify-Email header alongside the bearer token

Stays the same:

• Request body shape (comment.text, requestedAttributes, languages)
• Response structure (attributeScores, summaryScore, spanScores)
• Your thresholds and downstream logic
• Your suggestScore calls (different path, same shape)

Before you start

Sign up and get your API key from the dashboard. Add two environment variables:

RESPECTIFY_API_KEY=your-api-key
RESPECTIFY_EMAIL=you@example.com

TypeScript / JavaScript

Before (Perspective)

const response = await fetch(
  `https://commentanalyzer.googleapis.com/v1alpha1/comments:analyze?key=${process.env.PERSPECTIVE_API_KEY}`,
  {
    method: "POST",
    headers: { "Content-Type": "application/json" },
    body: JSON.stringify({
      comment: { text: commentText },
      requestedAttributes: { TOXICITY: {}, INSULT: {} },
    }),
  }
);

const data = await response.json();
const toxicityScore = data.attributeScores.TOXICITY.summaryScore.value;

After (Respectify)

const response = await fetch("https://app.respectify.org/v0.2/perspective-compat/analyse", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    Authorization: `Bearer ${process.env.RESPECTIFY_API_KEY}`,
    "Respectify-Email": process.env.RESPECTIFY_EMAIL,
  },
  body: JSON.stringify({
    comment: { text: commentText },
    requestedAttributes: { TOXICITY: {}, INSULT: {} },
  }),
});

const data = await response.json();
const toxicityScore = data.attributeScores.TOXICITY.summaryScore.value; // unchanged

Or use the TypeScript SDK to skip the raw fetch:

import { RespectifyClient } from "@respectify/client";

const client = new RespectifyClient({
  apiKey: process.env.RESPECTIFY_API_KEY,
  email: process.env.RESPECTIFY_EMAIL,
});

const result = await client.perspectiveCompat.analyse({
  comment: { text: commentText },
  requestedAttributes: { TOXICITY: {}, INSULT: {} },
});

const toxicityScore = result.attributeScores.TOXICITY.summaryScore.value;

Python

Before (Perspective)

import httpx

response = httpx.post(
    f"https://commentanalyzer.googleapis.com/v1alpha1/comments:analyze",
    params={"key": os.environ["PERSPECTIVE_API_KEY"]},
    json={
        "comment": {"text": comment_text},
        "requestedAttributes": {"TOXICITY": {}, "INSULT": {}},
    },
)

data = response.json()
toxicity_score = data["attributeScores"]["TOXICITY"]["summaryScore"]["value"]

After (Respectify)

import httpx

response = httpx.post(
    "https://app.respectify.org/v0.2/perspective-compat/analyse",
    headers={
        "Authorization": f"Bearer {os.environ['RESPECTIFY_API_KEY']}",
        "Respectify-Email": os.environ["RESPECTIFY_EMAIL"],
    },
    json={
        "comment": {"text": comment_text},
        "requestedAttributes": {"TOXICITY": {}, "INSULT": {}},
    },
)

data = response.json()
toxicity_score = data["attributeScores"]["TOXICITY"]["summaryScore"]["value"]  # unchanged

PHP

Before (Perspective)

$response = $client->post(
    "https://commentanalyzer.googleapis.com/v1alpha1/comments:analyze?key={$_ENV['PERSPECTIVE_API_KEY']}",
    [
        'json' => [
            'comment' => ['text' => $commentText],
            'requestedAttributes' => ['TOXICITY' => new stdClass(), 'INSULT' => new stdClass()],
        ],
    ]
);

$data = json_decode($response->getBody(), true);
$toxicityScore = $data['attributeScores']['TOXICITY']['summaryScore']['value'];

After (Respectify)

$response = $client->post(
    'https://app.respectify.org/v0.2/perspective-compat/analyse',
    [
        'headers' => [
            'Authorization' => 'Bearer ' . $_ENV['RESPECTIFY_API_KEY'],
            'Respectify-Email' => $_ENV['RESPECTIFY_EMAIL'],
        ],
        'json' => [
            'comment' => ['text' => $commentText],
            'requestedAttributes' => ['TOXICITY' => new stdClass(), 'INSULT' => new stdClass()],
        ],
    ]
);

$data = json_decode($response->getBody(), true);
$toxicityScore = $data['attributeScores']['TOXICITY']['summaryScore']['value']; // unchanged

Handling suggestScore

If your integration sends score corrections via Perspective’s suggestscore endpoint, the same pattern applies:

Note: feedback sent to suggestscore is retained by Respectify to improve the service. Don’t send content you aren’t comfortable storing.

Validate before going live

Run both endpoints in parallel on a sample of real comments and compare scores for your key attributes. Expect close alignment on TOXICITY and INSULT. Some Perspective attributes don’t have direct equivalents — check the attribute mapping docs if you use anything beyond the common set.

Once scores look right in staging, swap production. Your thresholds, queues, and admin tools need no changes.

What’s next

The compatibility endpoint gets you migrated. When you’re ready to go further, Respectify can stop bad comments before they post — and explain to the commenter why, so they have a chance to revise. That’s configurable per site and doesn’t require touching the code you just migrated.

Get your API key · Migration overview · Full API docs