Skip to content

Test the API from the portal

Updated

What it does#

Client portal → Test API lets you extract a real document without writing any code: pick a key, pick a document type, drop a file, see the JSON. Then copy a working code snippet for the exact call you just made.

Use it to confirm DocXtract handles your documents before you build an integration.

Watch it first#

Extracting your first document from the portal

Loads from YouTube only when you press play.

The written steps below cover the same ground — use whichever suits you. Every command and setting shown in the video is also written here, because text is searchable and copyable and video is neither.

Steps#

  1. Sign in at https://app.docxtract.io/ and open Test API
  2. Select API Key — only your active keys appear, with remaining quota shown
  3. Document Type — populated from that key's permitted models, so you cannot pick a type the key would reject
  4. Drop a PDF, JPG, or PNG (max 10 MB)
  5. Extract Data

Results show the extracted fields, processing stats, and the raw JSON.

Note
Test runs are real API calls against your own key and consume credits, one per page. They also count toward your rate limit.

Large PDFs work here too#

A PDF over 3 pages is split by the API and needs the multi-page flow. The portal runs that whole loop for you and shows the combined result, so what you see matches what an SDK would return.

Very large documents are the exception: past roughly 60 parts the portal declines and points you at the API or an SDK, since a browser-driven run is the wrong tool for a job that size.

Copying the code#

After a successful run, Make this call from your code shows the same call in cURL, PHP, Python, and Node — generated from the URL and options actually used, so it cannot drift from what ran.

Warning
Snippets contain YOUR_API_KEY or an environment variable, never your real key. Copy the key separately from Settings, and keep it in an environment variable rather than in source control.

The PHP snippet uses the PHP SDK, which collapses the multi-page flow into a single extract() call. The raw-HTTP snippets show the synchronous path only — handling the 202 branch yourself is on you.

Your key stays server-side#

The portal never puts your API key in the page or sends it from your browser. The upload goes to a portal endpoint that resolves the key server-side, scoped to your own account, and calls the API on your behalf.

Related