In CVAT versions 2.2.0 through 2.54.0, an attacker who can create a maliciously crafted label or upload an SVG image containing executable script can cause that script to run when a victim user edits the label, views a shape that references the label, or uploads a skeleton SVG. The injected JavaScript executes within the victim’s UI session, granting the attacker temporary access to all resources that the victim user can access.

The affected product is CVAT, developed by cvat-ai. All releases from 2.2.0 up to and including 2.54.0 are vulnerable. Version 2.55.0 onward contains the fix.

The vulnerability carries a CVSS score of 8.6, indicating high severity, but the EPSS score is below 1 %, suggesting a low likelihood of exploitation at present. The issue is not listed in the CISA KEV catalog. Exploitation requires the attacker to create a malicious label or upload a crafted SVG and then persuade or trick a victim user to edit or view the associated content. Therefore the vector is most likely user‑interaction‑based XSS delivered via business or project content.