# UNK_MassTraction: China-Aligned Actor Exploits Roundcube CVE-2024-42009 & CVE-2025-49113 to Deploy IceCube Stealer and VShell Against University Physics Departments

> Since May 2026, Proofpoint has tracked UNK_MassTraction, a suspected China-aligned espionage threat cluster, exploiting an XSS flaw in Roundcube webmail (CVE-2024-42009) to run the JavaScript stealer IceCube, then chaining a Roundcube PHP deserialization flaw (CVE-2025-49113) to drop the SquareShell webshell and the VShell Go-based backdoor. Targeting is narrowly focused on physics and engineering department administrators and professors at US and Canadian universities with national-security-relevant astrophysics and particle physics research.

- **Published:** 2026-07-13T00:00:00Z
- **Last reviewed:** 2026-07-13T00:00:00Z
- **Canonical:** https://intel.threadlinqs.com/threat/TL-2026-1259
- **Card image:** https://intel.threadlinqs.com/og/threat/threat-intel-unk-masstraction-icecube-vshell-chain-tl-2026-1259.webp?v=de5b1f3d (Threadlinqs threat intel: UNK_MassTraction Roundcube Chain: Detection & IOCs (TL-2026-1259) — High severity, 18 IOCs, 9 detection rules.)
- **ID:** TL-2026-1259
- **Severity:** HIGH (CVSS 9.9)
- **Category:** CAMPAIGN
- **Status:** ACTIVE
- **Actor:** UNK_MassTraction (China)
- **Detections:** 9 (1 shown free below) · **IOCs:** 18 (3 shown free below)
- **CVEs:** CVE-2024-42009, CVE-2025-49113

## Description

UNK_MassTraction is a suspected China-aligned espionage threat cluster first tracked by Proofpoint beginning in May 2026. The campaign opens with phishing emails sent from compromised third-party mailboxes (e.g. jpcontreras@newfield.cl) and spoofed, DMARC-weak domains, crafted to look like generic marketing/spam so recipients do not scrutinize them. The messages carry a specially crafted HTML body that abuses insufficient sanitization in Roundcube Webmail's mail-display function: an animation-event handler (onanimationstart) executes a decimal-encoded JavaScript loader the moment the message is opened in a vulnerable Roundcube client, tracked as CVE-2024-42009 (CVSS 9.3, CWE-79 stored/reflected XSS).

That loader delivers IceCube, a purpose-built JavaScript credential and session stealer that runs inside the victim's authenticated Roundcube session. IceCube performs DOM traversal to escape iframes, harvests usernames/passwords and any 2FA material entered into the page, steals session and CSRF cookies/tokens, fingerprints the browser (language, screen size, form data), and sets deferred triggers on the logout button and page-close events so it can clean up evidence and destroy forensic traces of the session before the victim notices. Code review of IceCube samples shows verbose, LLM-generated code comments and iterative "phase" markers, suggesting AI-assisted malware development, plus resilient retry/fallback logic for reliability across variable network conditions.

With stolen session material, the actor pivots to a second, more serious Roundcube flaw: CVE-2025-49113 (CVSS 9.9, CWE-502), an insufficiently validated `_from` parameter in `program/actions/settings/upload.php` that allows an authenticated user to trigger PHP object deserialization and reach a gadget chain culminating in `__destruct()`-driven command execution. The actor sends a crafted HTTP POST containing serialized PHP payload data (with a valid CSRF token obtained via the IceCube session hijack) to instantiate the gadget chain and execute arbitrary shell commands on the mail server itself — moving from a client-side webmail compromise to full server-side RCE on the mailserver host.

Server-side access is used to drop SquareShell, a small PHP webshell written to `plugins/newmail_notifier/mail_preview.php` inside the Roundcube plugin directory and timestomped to match the legitimate plugin's modification time to blend in with routine file activity. SquareShell exposes `system`, `passthru`, `exec`, `shell_exec`, `assert`, and `popen` execution primitives to give the operator durable, remotely reachable command execution independent of any further Roundcube exploitation.

Beginning in June 2026 the group introduced a fallback deployment mechanism: rather than relying solely on the PHP webshell, compromised servers are used to pull down and in-memory-load VShell, a mature Go-based, multi-platform (Windows/Linux/macOS) backdoor previously documented in numerous China-nexus campaigns (UNC5174/Uteus, UAT-8302, CL-STA-0048, Earth Lamia's "Operation DRAGONCLONE"). The loader downloaded onto UNK_MassTraction victims is architecture-dependent, checks `/tmp/log_de.log` to avoid duplicate execution, wraps itself with `nohup`, and masquerades in the process table as `[kworker/0:2]` — a well-known VShell evasion signature — to blend into normal Linux kernel worker-thread noise. Once running, VShell gives the operator an interactive shell, port-forwarding, and network-pivoting capability entirely in memory, without touching disk, complicating file-based detection and incident response on university-owned Linux mail infrastructure.

Targeting is unusually narrow and deliberate: rather than broad opportunistic phishing, UNK_MassTraction concentrates on physics and engineering department administrators and professors at US and Canadian universities, prioritizing institutions with astrophysics and particle physics programs carrying national-security relevance (e.g. dual-use research, DOE/DoD-funded labs, ITAR-adjacent instrumentation). Attribution to a China-aligned cluster rests on: access to covert VPS infrastructure shared with other known China-nexus operators; Chinese-language artifacts present in phishing email HTML source; and VShell's consistent association with Chinese state-linked toolkits and operators (UNC5174, UAT-8302, Earth Lamia, CL-STA-0048) across unrelated campaigns and targets. Proofpoint scanned relevant internet address space to identify additional SquareShell-infected Roundcube servers beyond the observed sample set and coordinated victim notification with government and industry partners.

Both exploited CVEs were independently confirmed under active exploitation by CISA, which added CVE-2025-49113 (alongside CVE-2025-68461) to the Known Exploited Vulnerabilities catalog on 2026-02-20 with a federal remediation deadline of 2026-03-13; a commercial exploit for CVE-2025-49113 was reportedly for sale within 48 hours of public disclosure. CVE-2024-42009 has separately been observed exploited by unrelated actors including APT28 and Winter Vivern, underscoring Roundcube's status as a recurring high-value target due to its wide deployment and historically slow organizational patch cadence.

## MITRE ATT&CK

- T1589 Gather Victim Identity Information
- T1586 Compromise Accounts
- T1583 Acquire Infrastructure
- T1566 Phishing
- T1190 Exploit Public-Facing Application
- T1059 Command and Scripting Interpreter
- T1204 User Execution
- T1505 Server Software Component
- T1068 Exploitation for Privilege Escalation
- T1036 Masquerading
- T1070 Indicator Removal
- T1620 Reflective Code Loading
- T1140 Deobfuscate/Decode Files or Information
- T1539 Steal Web Session Cookie
- T1555 Credentials from Password Stores
- T1111 Multi-Factor Authentication Interception
- T1082 System Information Discovery
- T1005 Data from Local System
- T1056 Input Capture
- T1571 Non-Standard Port
- T1071 Application Layer Protocol
- T1090 Proxy
- T1550 Use Alternate Authentication Material
- T1531 Account Access Removal

## Free detection (Splunk SPL, Microsoft KQL, Sigma)

**Roundcube onanimationstart XSS Loader Delivery (CVE-2024-42009)** (`TL-2026-1259-SPL-001`)

Detects proxy/web-log traffic to a Roundcube webmail endpoint carrying an onanimationstart CSS animation-event handler paired with decimal-encoded JavaScript — the IceCube stealer loader trigger for CVE-2024-42009.

- **Languages:** Splunk SPL, Microsoft KQL, Sigma
- **ATT&CK:** T1190, T1059.007
- **Data sources:** web_proxy_logs, roundcube_access_logs
- **Severity:** critical
- **Confidence:** high

### Splunk SPL

```spl
index=web sourcetype="proxy" (uri_path="*roundcube*" OR http_referrer="*roundcube*")
| regex uri_query="(?i)onanimationstart"
| eval decimal_encoded=if(match(uri_query, "(?:&#\d{2,4};){5,}"), 1, 0)
| where decimal_encoded=1
| stats count min(_time) as first_seen max(_time) as last_seen values(dest_ip) as dest_ips by src_ip, uri_path
| where count > 0
```

### Microsoft KQL

```kql
let lookback = 24h;
let vshell_ports = dynamic([23088, 8080]);
WebProxyLog
| where TimeGenerated > ago(lookback)
| where UrlPath has "roundcube" or UrlOriginal has "roundcube"
| where UrlQuery has "onanimationstart"
| extend DecimalEncodedHits = countof(UrlQuery, @"&#\d{2,4};", "regex")
| where DecimalEncodedHits >= 5
| project TimeGenerated, SrcIpAddr, UrlOriginal, DecimalEncodedHits
| summarize count(), min(TimeGenerated), max(TimeGenerated) by SrcIpAddr
```

### Sigma

```yaml
title: Roundcube onanimationstart XSS Loader Pattern (CVE-2024-42009)
id: 6d4e8f21-9a1b-4c3d-8e7f-1b2c3d4e5f6a
status: experimental
description: Detects an animation-event handler (onanimationstart) paired with decimal-encoded JavaScript in a Roundcube webmail HTTP request, indicative of IceCube stealer loader delivery via CVE-2024-42009.
references:
    - TL-2026-1259
    - https://www.proofpoint.com/us/blog/threat-insight/one-email-closer-edge-unkmasstraction-physics-exploitation
author: Threadlinqs Team
date: 2026/07/13
tags:
    - attack.initial_access
    - attack.t1190
    - attack.execution
    - attack.t1059.007
    - cve.2024.42009
logsource:
    category: proxy
    product: web
detection:
    selection_path:
        c-uri|contains: 'roundcube'
    selection_trigger:
        c-uri|contains: 'onanimationstart'
    condition: selection_path and selection_trigger
falsepositives:
    - Legitimate CSS animation debugging traffic (rare on production webmail)
level: high
```

**Known false positives:**

- Legitimate use of onanimationstart in CSS/JS testing tools rarely observed against production webmail endpoints

**Tuning notes:** Restrict source to internet-facing Roundcube hosts; tune the decimal-encoding threshold (currently >=5 sequential &#NNN; entities) down if loader variants use shorter payloads.

### 8 more detection rules (Splunk SPL, Microsoft KQL, Sigma), unlock with TL Creds

- CVE-2025-49113 PHP Deserialization RCE Attempt via upload.php (locked: unlock with TL Creds)
- SquareShell Webshell Command Execution Primitive Invocation (locked: unlock with TL Creds)
- VShell kworker Process Masquerade with Non-standard-Port C2 Beacon (locked: unlock with TL Creds)
- IceCube In-Session Credential and Web-Session-Cookie Theft Pattern (locked: unlock with TL Creds)
- VShell Duplicate-Execution Marker File Creation on Linux Mailserver (locked: unlock with TL Creds)
- IceCube Loader Masquerading as jQuery from sslip.io Infrastructure (locked: unlock with TL Creds)
- SquareShell Webshell Drop at Roundcube Plugin Path (CVE-2025-49113) (locked: unlock with TL Creds)
- VShell Fileless C2 Egress from Roundcube Mailserver Host (locked: unlock with TL Creds)

Sign up free to unlock the detection rules: https://threadlinqs.com/?modal=signup

## Sample indicators (3 of 18)

Defanged, TLP:CLEAR.

- IP address: `45.150.109[.]151`
- File hash: `a02f124c5ce4180bd130a62ee03262f399c33491de3aed36e0b15155ae4926c0`
- IP address: `194.213.18[.]133`

Full set of 18 indicators unlocks with TL Creds: https://threadlinqs.com/tl-creds

## Sources

- [One Email Closer to the Edge: UNK_MassTraction & the Physics of Exploitation](https://www.proofpoint.com/us/blog/threat-insight/one-email-closer-edge-unkmasstraction-physics-exploitation)
- [Malpedia library entry (js.icecube)](https://malpedia.caad.fkie.fraunhofer.de/library/e498402e-a08c-402e-88a6-fcb63f8f34d4/)
- [Roundcube security updates 1.6.8 and 1.5.8](https://roundcube.net/news/2024/08/04/security-updates-1.6.8-and-1.5.8)
- [Roundcube security updates 1.6.11 and 1.5.10](https://roundcube.net/news/2025/06/01/security-updates-1.6.11-and-1.5.10)
- [Government emails at risk: critical cross-site scripting vulnerability in Roundcube Webmail](https://sonarsource.com/blog/government-emails-at-risk-critical-cross-site-scripting-vulnerability-in-roundcube-webmail/)
- [CISA Known Exploited Vulnerabilities Catalog](https://www.cisa.gov/known-exploited-vulnerabilities-catalog)
- [CISA Adds Two Actively Exploited Roundcube Flaws to KEV Catalog](https://thehackernews.com/2026/02/cisa-adds-two-actively-exploited.html)
- [CISA: Recently patched RoundCube flaws now exploited in attacks](https://www.bleepingcomputer.com/news/security/cisa-recently-patched-roundcube-flaws-now-exploited-in-attacks/)
- [The Silent, Fileless Threat of VShell](https://www.trellix.com/blogs/research/the-silent-fileless-threat-of-vshell/)
- [VShell (Malware Family)](https://malpedia.caad.fkie.fraunhofer.de/details/win.vshell)
- [China-Linked UAT-8302 Targets Governments Using Shared APT Malware Across Regions](https://thehackernews.com/2026/05/china-linked-uat-8302-targets.html)
- [VShell and SparkRAT Observed in Exploitation of BeyondTrust Critical Vulnerability (CVE-2026-1731)](https://unit42.paloaltonetworks.com/beyondtrust-cve-2026-1731/)
- [Decoding VShell: Insights into a Chinese-Language Cyber Espionage Tool](https://blog.nviso.eu/wp-content/uploads/2025/11/VShell.pdf)
- [UAT-8302 and its box full of malware](https://blog.talosintelligence.com/uat-8302/)
- [UNC5174's evolution in China's ongoing cyber warfare: From SNOWLIGHT to VShell](https://www.sysdig.com/blog/unc5174-chinese-threat-actor-vshell)

## Full data

The other 8 detection rules (Splunk SPL / Microsoft KQL / Sigma) unlock with TL Creds; a free account includes daily creds: https://threadlinqs.com/?modal=signup
Indicator values unlock with TL Creds, at a lower price on Red and Purple; AI agents can query rules and IOCs through the Threadlinqs MCP server (Purple): https://intel.threadlinqs.com/mcp

Canonical: https://intel.threadlinqs.com/threat/TL-2026-1259
