testosaurusdocs
Browse docs
On this page

Session replay

Testo waiting

Session replay records recent page activity so a report can show what led to the bug alongside the description. It is strictly opt-in: recording starts when replay consent is granted. Withdrawing consent stops recording and clears the unsubmitted replay buffer. See Privacy & consent.

What gets recorded

  • DOM changes — element insertions, moves, and visibility so the playback reconstructs the page.
  • Interactions — clicks, input focus, scroll position, and route changes.

The feedback widget itself is never recorded, so the tester's message and the widget's own interface stay out of the playback.

Masking

Input, textarea and select values and text typed in rich-text editors (any contenteditable region) are masked. You see that a field was filled, not what was typed. A masked select shows its first option rather than the one the visitor chose; nothing the visitor picked is recorded. Hidden inputs (<input type="hidden">), which often carry CSRF tokens or account ids, are left out. Other visible page text is recorded; add data-testosaurus-mask to mask an element's text or data-testosaurus-block to leave it out.

  • data-testosaurus-mask — the element and everything inside it keep their layout, but their text is replaced. Use it for names, email addresses, account numbers and other personal data shown on the page. Attribute values such as title, alt and aria-label are still recorded, so block elements whose attributes hold personal data.
  • data-testosaurus-block — the element is replaced by an empty placeholder of the same size. Use it for whole regions that must never leave the browser, such as a payment form or an embedded document.
  • The recorder's own rr-mask and rr-block classes work the same way.
<span data-testosaurus-mask>jane.doe@example.com</span>
<section data-testosaurus-block>…card details…</section>

The same attributes apply to screenshots, so one set of markers covers both captures. The page address stored with each snapshot follows the same URL redaction as the report context, so tokens in the query or fragment don't reach the replay.

How the buffer works

  • Full snapshots: the recorder takes a fresh full page snapshot every 60 seconds and after every 3000 recorded events. The buffer keeps the previous and the current snapshot with the events that follow them, so every replay starts from a complete page.
  • Bounded size: the older snapshot is dropped once the buffer passes about 8 MB or 6000 events.
  • Upload on submit: the buffered events can accompany the report when replay consent is granted. A buffer above the 10 MB upload limit sends only the latest snapshot and its events. After an upload the buffer is cleared and, if consent is still granted, a fresh recording starts. If an upload fails, the buffer is trimmed to the latest snapshot so the next report can still carry a replay.
  • Nothing survives unload: when the page closes without a submit, the buffer is dropped.

Playback

Reports with replay show a player in the dashboard's detail view. It fits the evidence panel in light and dark appearance; use the full-screen control for a larger view. Scrub through it next to the distilled summary to see the exact moment things went wrong.

Something unclear?Let us know