CVE-2026-39040: BharatMLStack XSS Vulnerability

CVE-2026-39040 Overview

CVE-2026-39040 is a Cross-Site Scripting (XSS) vulnerability in BharatMLStack versions up to and including 1.3.0. The flaw resides in the Trufflebox UI component (trufflebox-ui), specifically within the ExpressionViewModal.jsx file. BharatMLStack is an open-source machine learning platform maintained by Meesho. Attackers can inject malicious JavaScript that executes in the browser context of users interacting with the affected modal. Successful exploitation can lead to session hijacking, credential theft, and unauthorized actions performed on behalf of authenticated users of the Trufflebox interface.

Critical Impact
An attacker can execute arbitrary JavaScript in the browsers of Trufflebox UI users, enabling session theft and account takeover within the BharatMLStack management interface.

Affected Products

BharatMLStack versions up to and including 1.3.0
Trufflebox UI component (trufflebox-ui)
ExpressionViewModal.jsx React component

Discovery Timeline

2026-09-15 - CVE-2026-39040 published to NVD
2026-09-15 - Last updated in NVD database

Technical Details for CVE-2026-39040

Vulnerability Analysis

The vulnerability is a Cross-Site Scripting (XSS) flaw [CWE-79] within the Trufflebox UI, a React-based management interface for BharatMLStack. The defect exists in the ExpressionViewModal.jsx component, which is used to render expression data inside a modal dialog.

The component fails to properly sanitize or encode user-controlled input before rendering it in the Document Object Model (DOM). When an attacker-supplied expression string is displayed, embedded HTML or JavaScript payloads execute in the victim's browser context. This behavior is consistent with client-side XSS caused by unsafe rendering patterns in React, such as use of dangerouslySetInnerHTML or unescaped interpolation of raw HTML.

Because Trufflebox UI is used to manage machine learning workflows and features, compromised sessions can grant attackers access to sensitive ML feature data, model metadata, and platform configuration.

Root Cause

The root cause is missing output encoding of untrusted expression data before it is inserted into the rendered modal view. Input that should be treated as inert text is instead interpreted as HTML markup, allowing script tags and event handlers to execute in the user's browser.

Attack Vector

An attacker with the ability to submit or influence expression content within BharatMLStack stores a malicious payload. When a legitimate Trufflebox user opens the ExpressionViewModal view containing that payload, the browser executes the attacker's JavaScript. The payload runs with the privileges of the viewing user, enabling theft of authentication tokens stored in cookies or localStorage, forged API requests, and lateral movement within the platform. Refer to the GitHub CVE Analysis Writeup for detailed exploitation notes.

Detection Methods for CVE-2026-39040

Indicators of Compromise

Expression records containing HTML tags such as <script>, <img onerror=...>, or <svg onload=...> in BharatMLStack data stores.
Unexpected outbound requests from Trufflebox user browsers to attacker-controlled domains shortly after opening expression modals.
Session tokens or cookies for Trufflebox accounts appearing in web server logs from unfamiliar IP addresses.

Detection Strategies

Review stored expression fields for HTML or JavaScript content that should not exist in structured expression data.
Enable and monitor Content Security Policy (CSP) violation reports from the Trufflebox UI to catch inline script execution attempts.
Inspect browser console errors and network telemetry from workstations that access the Trufflebox management console.

Monitoring Recommendations

Log and alert on write operations to expression endpoints that contain angle brackets or common XSS payload signatures.
Correlate Trufflebox authentication events with anomalous API activity from the same session tokens.
Track version metadata for deployed BharatMLStack instances to identify hosts still running 1.3.0 or earlier.

How to Mitigate CVE-2026-39040

Immediate Actions Required

Restrict access to the Trufflebox UI to trusted administrative users and networks until a patched release is deployed.
Audit all existing expression records for embedded HTML or script content and remove any suspicious entries.
Rotate session tokens and credentials for any user who accessed the affected modal during the exposure window.

Patch Information

No fixed version is listed in the NVD entry at the time of publication. Monitor the BharatMLStack GitHub repository for a release that supersedes 1.3.0 and includes proper output encoding in ExpressionViewModal.jsx.

Workarounds

Deploy a strict Content Security Policy that disallows inline scripts and restricts script sources for the Trufflebox UI.
Place the Trufflebox interface behind a web application firewall (WAF) with rules that block common XSS payloads on write endpoints.
Sanitize expression input at the API layer using an allowlist that rejects HTML markup before persisting data.

bash

# Example Content Security Policy header for the Trufflebox UI reverse proxy
add_header Content-Security-Policy "default-src 'self'; script-src 'self'; object-src 'none'; base-uri 'self'; frame-ancestors 'none'";