Description

In JetBrains Datalore before 2026.4 stored XSS in notebook charts was possible via Vega editor link

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "5.4",
  "pubDate": "2026-11-06T11:17:07.310Z",
  "pubdate": "2026-11-06T11:17:07.310Z",
  "executiveSummary": "A Stored Cross-Site Scripting (XSS) vulnerability exists in JetBrains Datalore versions prior to 2026.4.\nThe vulnerability resides within the Vega editor link functionality, allowing an attacker to inject malicious scripts into notebook charts.\nSuccessful exploitation enables an attacker to execute arbitrary JavaScript within the context of a victim's browser session when they view the affected chart.\nThis impact includes session hijacking, credential theft, unauthorized actions performed on behalf of the user, and potential data exfiltration from the Datalore environment.\nThe vulnerability affects Datalore's chart rendering and input sanitization mechanisms.\nExploitation requires the attacker to have the ability to create or edit notebook charts containing a Vega editor link, which is then persisted in the application's storage.\nThis poses a significant risk to collaborative environments where multiple users share or view notebook charts.",
  "technicalDetails": "The vulnerability is a Stored XSS flaw stemming from insufficient input sanitization and output encoding within the Vega editor link component of JetBrains Datalore.\nVega is a visualization grammar that allows for the creation of complex interactive charts. In Datalore, these charts can contain links defined within the Vega specification.\nThe root cause of this vulnerability is the improper validation or sanitization of input data provided via the Vega editor interface when defining hyperlinked elements within a chart.\nAn attacker can craft a malicious Vega specification where a hyperlink attribute contains a 'javascript:' URI scheme or other payload vectors designed to execute code upon user interaction or rendering.\nBecause the application stores these specifications within the notebook data, the malicious payload is persisted on the server.\nThe attack flow proceeds as follows: First, an authenticated attacker with permissions to edit notebooks creates or modifies a chart using the Vega editor. The attacker injects a malicious payload into a link field within the Vega JSON structure. Second, the server accepts and stores this malicious chart specification as part of the notebook. Third, when an unsuspecting user views or interacts with the rendered chart in the Datalore UI, the application renders the embedded link without adequate sanitization.\nConsequently, the browser interprets the injected JavaScript payload as legitimate code belonging to the Datalore origin. This execution occurs within the security context of the victim's session, granting the attacker access to the document object model (DOM), cookies, and other sensitive session storage accessible to the Datalore domain.\nThe malicious script could then perform various post-exploitation activities, such as exfiltrating data, performing unauthorized API calls, or redirecting the user to malicious sites.\nThe vulnerability persists until the chart is updated with a sanitized version or the affected notebook is removed, as the payload is stored server-side and served to every user who accesses the notebook.\nThis flaw underscores a failure in enforcing a strict Content Security Policy (CSP) or improper usage of sanitization libraries when rendering user-supplied visualization metadata."
}