CVE-2026-121928

RAGFlow’s document image API can expose readers to cross-site scripting

InfiniFlow · RAGFlow document images API

First documented disclosure · Review checkpoint

Summary

VulDB identifies a browser script-injection weakness in how the image API serves content.

Severity and exploitation evidence

Medium · 5.1/10 · CVSS 4.0. Severity source.

Exploitation not established: The CNA reports a public exploit. A malicious attack or affected user is not established in the reviewed sources.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

The earlier route could select an image media type from an object’s extension. The named fix tests the actual bytes and rejects unrecognized image content.

Conditions required

An attacker can influence content served as an image, and a user must load it in a browser context where it can execute. The CNA does not supply a complete browser reproduction.

Potential impact

Script could act in the viewing origin. The exact privileges and data available to such a script were not independently demonstrated.

Affected versions and configurations

VulDB lists versions through 0.27.2.

Fix and mitigation status

VulDB identifies 0.27.3. The named change checks media signatures in addition to checking document ownership.

Practical next steps

Update to a release containing the fix. Verify that the image route rejects active content even if the filename has an image extension.

Operating lesson

Check the bytes served to the browser separately from the user’s permission to read the object.

Evidence and limits

September 18 is the public proposal for the named image route change. Earlier issue 14763 concerned unauthorized object reads. VulDB assigned the XSS record November 4. The earlier access-control report does not by itself demonstrate XSS.

Sources