Dev.to Security 🔐 Cybersecurity 👁 0 📖 4 min read

The Convertr Skimmer's Evidence Ladder: From 22 Page Matches to Confirmed Data Theft

The Convertr Skimmer's Evidence Ladder: From 22 Page Matches to Confirmed Data Theft The Convertr incident produced several kinds of numbers. A public query for the exact tracker hostname returned 22 page observations.

The Convertr Skimmer's Evidence Ladder: From 22 Page Matches to Confirmed Data Theft

The Convertr incident produced several kinds of numbers. A public query for the exact tracker hostname returned 22 page observations. A Plausible product query returned 285,765. Sansec also described the skimmer's behavior and exfiltration endpoint. Those numbers and facts do not carry the same evidentiary weight.

That difference matters because a security article can easily turn a public match into an affected-merchant count or a victim estimate. The Convertr record does not support either conversion. It supports a smaller, useful chain of evidence.

The narrowest observation

At 14:50:18 UTC on November 18, 2026, an all-scope ZoomEye search for http.body="analytics.convertr.app.br" returned 22 public observations [5]. The query identifies pages whose indexed HTML body contains the tracker hostname. It does not prove that the page executed the script, that the associated store is a Convertr customer or that the script was malicious at collection time.

The exact field and scope are part of the claim. A different query would produce a different number. A page-title search, a hostname query or a broad Plausible search would answer another question. Mixing the results without labels would make the evidence look stronger than it is.

The product baseline

The query for http.body="plausible.io" returned 285,765 observations [6]. This shows that references to the standard Plausible domain are widespread in the index. It can help an analyst understand the product context. It cannot establish Convertr usage because Convertr ran a self-hosted tracker on a different hostname.

The gap between 22 and 285,765 is a warning against scope drift. A responder looking for Convertr dependencies should first inspect the exact host. A broad product count may be useful later, but it should not replace the incident-specific observation.

The endpoint baseline

A query for http.body="/storybook" returned 70,829 observations [7]. Plausible's advisory says the exposed endpoint is part of the affected condition for Community Edition versions before 3.2.1 [2][3]. The query string alone cannot establish the application, the version or the response.

The endpoint query is therefore a discovery aid. It can help form a candidate list, but the analyst needs a direct response and service evidence before calling an instance affected.

The incident evidence

Sansec reported that the tracker was modified, that it activated on checkout.convertr pages and that payment fields were sent to 1x0.is [1]. That is much stronger than a page-body match. It includes a code-level observation and a described exfiltration destination.

The report also says the provider removed the skimmer and upgraded Plausible. It says the tracker was clean at the later check [1]. The initial access remained suspected. The researchers did not have server logs and could not confirm that the /storybook flaw caused the modification.

Build a defensible evidence ladder

The first level is a public string match. The second is a rendered checkout that loads the tracker. The third is a confirmed ownership relationship between the tracker and the merchant. The fourth is a copied suspicious response with skimming code. The fifth is observed transmission of payment fields to an attacker endpoint. The sixth is provider, bank or transaction evidence that customers were harmed.

Every level supports a different sentence. "A page contained the string" belongs to level one. "The checkout loaded the modified tracker" belongs to level two or four, depending on what was captured. "The incident stole N cards" belongs near level six and needs evidence that this article does not have.

Use the ladder in practice

Start with the exact hostname and preserve the query details. Move to a direct browser capture. Verify the merchant relationship. Compare the tracker body with an approved release. Review Plausible logs and file hashes where the provider permits. Coordinate with payment providers before making customer-impact claims.

This workflow does not make the incident smaller. It makes the response more precise. It shows which assets need investigation, which code needs containment and which claims need more evidence.

The 22 page observations are useful because they are bounded. The 285,765 Plausible observations are useful because they establish product context. The Sansec analysis is useful because it describes malicious behavior. Keeping those statements separate is what turns a collection of findings into a credible investigation.

References

[1] Sansec, "Notorious Brazilian Magecart group hijacks Convertr checkouts via outdated Plausible install." https://sansec.io/research/convertr-plausible-supply-chain-attack

[2] Plausible, "Unauthenticated remote code execution via /storybook endpoint." https://github.com/plausible/analytics/security/advisories/GHSA-mhcv-h7gf-57cf

[3] Plausible, "Plausible Analytics releases." https://github.com/plausible/analytics/releases

[4] BleepingComputer, "Hackers hijack Convertr analytics script to steal payment card data." https://www.bleepingcomputer.com/news/security/hackers-hijack-convertr-analytics-script-to-steal-payment-card-data/

[5] ZoomEye observation for http.body="analytics.convertr.app.br", collected November 18, 2026. https://www.zoomeye.ai/searchResult?q=aHR0cC5ib2R5PSJhbmFseXRpY3MuY29udmVydHIuYXBwLmJyIg%3D%3D

[6] ZoomEye observation for http.body="plausible.io", collected November 18, 2026. https://www.zoomeye.ai/searchResult?q=aHR0cC5ib2R5PSJwbGF1c2libGUuaW8i

[7] ZoomEye observation for http.body="/storybook", collected November 18, 2026. https://www.zoomeye.ai/searchResult?q=aHR0cC5ib2R5PSIvc3Rvcnlib29rIg%3D%3D

📰 Read the original article on Dev.to Security

Originally published by Dev.to Security. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.