Guide

Finding a Website's Hidden API With Chrome DevTools (Step by Step)

Use the Network tab to locate the JSON behind any web app, read pagination and tokens, and judge feasibility.

Most "impossible to scrape" sites are actually the easiest — once you find the JSON they load themselves. This is the first thing Apex Automation Team does in every feasibility test, and you can do it in five minutes.

Step 1 — Open the Network tab

Press F12 → Network, tick "Preserve log", filter by Fetch/XHR. Reload the page.

Step 2 — Trigger the data

Scroll, click "next page", change a filter. Watch which requests fire. Responses that are JSON with the data you see on screen are your candidates.

Step 3 — Read the request

Check the URL parameters (page, offset, cursor, limit, query), the headers (authorization, x-api-key, csrf tokens, cookies) and the method. Right-click → "Copy as cURL" to reproduce it outside the browser.

Step 4 — Understand pagination and limits

Cursor-based pagination needs the "next" token from each response; offset-based can be parallelised carefully. Note rate limits from response headers or 429 errors.

Step 5 — Tokens and signatures

Some endpoints need a session cookie (log in once, reuse it), a short-lived bearer token (refresh it programmatically) or a request signature computed in JavaScript (reverse-engineer or drive a browser for that one call). This is where API extraction and reverse engineering earns its keep.

Step 6 — Decide

If the endpoint returns clean JSON with reasonable limits: direct extraction, fast and cheap. If everything is signed and rotating: hybrid — a browser establishes the session, the extractor uses it. If there is no endpoint at all: rendered scraping.

Read next: Web scraping vs API extraction. Or send us the URL and Apex Automation Team will do this analysis for free.

Quick checklist

  • Filter Network by Fetch/XHR and preserve the log
  • Trigger pagination and filters to reveal endpoints
  • Copy the request as cURL and replay it
  • Note pagination type, limits and required headers
  • Check whether tokens expire or are signed

Frequently asked

What if the site uses GraphQL?

GraphQL endpoints are excellent for extraction: one endpoint, structured queries, predictable pagination.

Is using a hidden API allowed?

It depends on the site's terms and the data. We collect only public data and respect rate limits; unsure cases get a frank risk note in the feasibility test.