Skip to navigation

Threat reportPhishingTL-2026-2490

Smishing Triad "Outsider" Operator: JWR Phishing Kit's AES-256-CTR WebSocket Exfiltration Cockpit

highACTIVE

Group-IB details the JWR phishing kit used by "Outsider" (operator alias ChenLun, 局外人), a Chinese-language sub-cluster within the Smishing Triad PhaaS ecosystem. JWR is a Vue 2 single-page application that harvests a ~70-field victim data model over disposable landing domains and exfiltrates it in real time over an AES-256-CTR-encrypted WebSocket channel to operator consoles, enabling live manipulation of victims across 32 controllable phishing pages.

Key facts

IdentifiersTL-2026-2490
Also known asJWR phishing kit, Outsider PaaS kit, 局外人
SeverityHIGH
Affected productsMultiple impersonated brands (regional banks, brokerages, telecom carriers, postal/logistics operators, government toll and fine agencies, PayPal) SMS-delivered brand-impersonation phishing pages served by the JWR/Outsider PaaS kit (no software vulnerability — this is a phishing-kit campaign, not a CVE)
AttributionOutsider (medium confidence)
MalwareJWR, Outsider, trsb.top
Toolsjavascript-obfuscator.io
MITRE ATT&CK techniquesT1027 Obfuscated Files or Information, T1027.013 Encrypted/Encoded File, T1041 Exfiltration Over C2 Channel, T1056.003 Web Portal Capture, T1071.001 Web Protocols, T1095 Non-Application Layer Protocol, T1111 Multi-Factor Authentication Interception, T1119 Automated Collection, T1140 Deobfuscate/Decode Files or Information, T1204.001 Malicious Link, T1217 Browser Information Discovery, T1497 Virtualization/Sandbox Evasion and 8 more in the attack flow
MotivationFINANCIAL
Target sectorsfinancial services, banking, brokerage, telecoms, postal and logistics, government toll collection, fines and penalties administration
Target regionsNorth America, Europe, Middle East, Asia-Pacific, Latin America
Detection rules9 (1 shown free)
Sources9 cited sources

Detection coverage

As of 2026-11-03, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-2490 across Splunk SPL, Microsoft KQL and Sigma, covering 38 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.

Free detection: JWR/Outsider Real-Time Victim Data Harvesting (Multi-Endpoint /api/open/* Co-Occurrence)

Detects a single client hitting four or more of the seven distinctive JWR PhaaS REST endpoints (addClick, getSyncSettings, addCvv, the_final_interface, updateStatus, getPendingInstruction, binLookup) within a short window — the behavioral fingerprint of active per-keystroke PII/card-data collection and BIN-based fraud triage against a live victim session, alongside the kit's automated 4-provider IP-geolocation rotation.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1056.003, T1119, T1027, T1614, T1217
  • Data sources: proxy_logs, web_gateway_logs, mde_device_network_events, waf_logs
  • Severity: critical
  • Confidence: high

Splunk SPL

index=proxy sourcetype="zscaler:nss:web" uri="/api/open/*"
| eval endpoint=case(
    match(uri,"addClick"),"addClick",
    match(uri,"getSyncSettings"),"getSyncSettings",
    match(uri,"addCvv"),"addCvv",
    match(uri,"the_final_interface"),"the_final_interface",
    match(uri,"updateStatus"),"updateStatus",
    match(uri,"getPendingInstruction"),"getPendingInstruction",
    match(uri,"binLookup"),"binLookup")
| where isnotnull(endpoint)
| bin _time span=5m
| stats dc(endpoint) as distinct_endpoints values(endpoint) as endpoints_seen count as request_count by _time, src_ip, dest_ip, dest_host
| where distinct_endpoints >= 4
| sort -distinct_endpoints

Microsoft KQL

DeviceNetworkEvents
| where Timestamp > ago(24h)
| where RemoteUrl has "/api/open/"
| extend Endpoint = extract(@"/api/open/(\w+)", 1, RemoteUrl)
| where Endpoint in ("addClick","getSyncSettings","addCvv","the_final_interface","updateStatus","getPendingInstruction","binLookup")
| summarize DistinctEndpoints = dcount(Endpoint), EndpointsSeen = make_set(Endpoint), RequestCount = count() by DeviceName, RemoteIP, bin(Timestamp, 5m)
| where DistinctEndpoints >= 4
| order by DistinctEndpoints desc

Sigma

title: JWR Phishing Kit REST Endpoint Access (Victim Data Harvesting)
id: 3c9d7e12-4a2f-4e91-9b6a-2f8d5c0e7a22
status: experimental
description: Detects requests to JWR PhaaS kit /api/open/* REST endpoints used for victim-arrival beaconing, config sync, per-keystroke PII/card exfiltration, funnel completion, status updates, and real-time BIN lookups. Endpoint names are unique to the JWR kit codebase; single-event matches are already high confidence, and co-occurrence of 4+ distinct endpoints from the same host within a short window (see companion SPL/KQL correlation searches) indicates an active harvesting session.
references:
  - https://www.group-ib.com/blog/smishing-triad-outsider-jwr/
author: Threadlinqs Team
date: 2026-09-14
tags:
  - attack.collection
  - attack.t1056.003
  - attack.t1119
  - attack.defense-evasion
  - attack.t1027
  - attack.discovery
  - attack.t1614
logsource:
  category: proxy
  product: proxy
detection:
  selection_addclick:
    c-uri|contains: '/api/open/addClick'
  selection_sync:
    c-uri|contains: '/api/open/getSyncSettings'
  selection_addcvv:
    c-uri|contains: '/api/open/addCvv'
  selection_final:
    c-uri|contains: '/api/open/the_final_interface'
  selection_status:
    c-uri|contains: '/api/open/updateStatus'
  selection_poll:
    c-uri|contains: '/api/open/getPendingInstruction'
  selection_bin:
    c-uri|contains: '/api/open/binLookup'
  condition: 1 of selection_*
falsepositives:
  - None expected; these endpoint path segments are unique to the JWR kit codebase and unlikely to collide with legitimate applications
level: high

Known false positives:

  • Automated security scanners or crawlers probing the same disposable domain could coincidentally hit multiple /api/open/* paths without a real victim session; cross-check against a valid JwrCvvForm/JWRCID session cookie or localStorage beacon if available

Tuning notes: Escalate to critical/P1 when distinct_endpoints includes 'addCvv' AND 'binLookup' together — this pairing indicates live card-data capture plus BIN-based fraud eligibility triage, the highest-value stage of the funnel; lower the 5-minute correlation window if the SIEM shows normal per-keystroke event volume causing false merges across unrelated sessions.

8 more detection rules (Splunk SPL, Microsoft KQL, Sigma) · unlock with TL Creds. Sign up free to unlock the detection rules or query them with the Threadlinqs MCP server (Purple)
  • JWR/Outsider Smishing Landing Page Access (Staged HTML + PayPal Funnel) · unlock with TL Creds
  • JWR/Outsider Real-Time Victim Data Harvesting (Multi-Endpoint /api/open/* Co-Occurrence) · unlock with TL Creds
  • JWR AES-256-CTR WebSocket Exfiltration Channel (Static Access-Token Path) · unlock with TL Creds
  • JWR Phishing Kit Staged HTML Page Access · unlock with TL Creds
  • JWR Phishing Kit REST Endpoint Access (Victim Data Harvesting) · unlock with TL Creds
  • JWR AES-256-CTR WebSocket Exfiltration Channel · unlock with TL Creds
  • JWR/Outsider Smishing Landing Page Access (Staged HTML + PayPal Funnel) · unlock with TL Creds
  • JWR AES-256-CTR WebSocket Exfiltration Channel (Static Access-Token Path) · unlock with TL Creds

Indicators of compromise

As of 2026-11-03, Threadlinqs Intelligence tracks 38 indicators of compromise for TL-2026-2490.

Sample indicators (1 of 38)

  • Domain: irs[.]gov-addpayment[.]info

Defanged, TLP:CLEAR. Full set of 38 indicators: unlock with TL Creds.

38 indicators of compromise · unlock with TL Creds. Compare plans

Technical analysis

JWR is a Phishing-as-a-Service kit built as a Vue 2 single-page application with a dedicated Web Worker handling encrypted communications. Its central `cvvform` object collects roughly 70 fields spanning identity/PII (name, address, DOB, SSN, passport/license/medical numbers), handheld and front/back document imagery for KYC defeat, card data (13 fields including BIN-enriched brand/issuer/country), up to three chained web-banking/brokerage credential slots, and a separate PayPal credential funnel, alongside operator metadata (operator_by, current_page, online_status, belong_to_template).

Exfiltration runs over a custom binary envelope — a 32-byte AES-256 key and 16-byte IV transmitted in clear, followed by AES-CTR ciphertext of UTF-8 JSON (key || iv || ciphertext, no key agreement) — pushed per keystroke over a WebSocket at `/webSocket/QT/<JWRCID>/khkjsahfjkwhakjlsdwdddddd88`, with a 2-second HTTP long-poll fallback (`/api/open/getPendingInstruction`) and six other `/api/open/*` REST endpoints for victim-arrival beaconing, config sync, exfiltration, funnel completion, status updates, and real-time BIN lookups. Group-IB notes the scheme is trivially decryptable with a few lines of Python — it only defeats naive logging, not an analyst.

The funnel itself is built from alphabetically staged HTML pages (a_ through z_) covering landing/login, password/PII/QR-verify, card capture, SMS/2FA/email-OTP capture, card PIN capture, app-confirmation stalling, and a full PayPal sub-funnel (bank/PayPal login through PIN/app confirmation). A 32-command `instructionConfig` map gives the human operator real-time control to redirect victims between stages, inject verification challenges, swap displayed QR codes, force card-declined responses, or flip into an `unattended_switch` autonomous mode that auto-approves whitelisted BINs passing a Luhn check and kills sessions that fail.

Attribution to the "Outsider" sub-cluster rests on converging technical fingerprints: Simplified Chinese operator-console strings (卡头/BIN, 无人值守/unattended, 需操作!), Hong Kong domain registration paired with U.S. cloud/CDN fronting, disposable landing domains active a median of two days or less, and the literal string "JWR" recurring across localStorage keys, CSS prefixes, and the per-victim ID format `JWRCVV-{epoch13}-{rand}-{rand}`; internal Web Worker strings additionally reference the codebase as part of a "trsb.top family." Group-IB separately profiles the same Outsider operator (alias ChenLun) and its AiTM-capable kit as having produced 100,000+ phishing pages across 54+ countries between December 2025 and May 2026 using 267+ templates, prompting a June 2026 Google civil suit and an FBI/Lumen Black Lotus Labs takedown effort ("Operation Ghost Hook") that the group survived, standing up 700+ new pages within a month.

JWR/Outsider sits inside the much larger Smishing Triad PhaaS ecosystem that Unit 42 and Silent Push have tracked since January 2024: 194,000+ malicious domains across 121+ countries, over two-thirds registered via Hong Kong registrar Dominet (HK) Limited, more than half resolving through Cloudflare (AS13335, notably the 104.21.0.0/16 block) or Chinese cloud ASNs, and cumulative fraud proceeds in the multi-billion-dollar range — including a Fortra-tracked fivefold year-over-year increase in brokerage-account targeting used for "ramp and dump" stock-manipulation schemes.

MITRE ATT&CK techniques by tactic:

Defense Evasion

T1027 Obfuscated Files or Information; T1027.013 Encrypted/Encoded File; T1140 Deobfuscate/Decode Files or Information; T1497 Virtualization/Sandbox Evasion; T1622 Debugger Evasion

Exfiltration

T1041 Exfiltration Over C2 Channel

Collection

T1056.003 Web Portal Capture; T1119 Automated Collection

Command and Control

T1071.001 Web Protocols; T1095 Non-Application Layer Protocol; T1571 Non-Standard Port; T1573.001 Symmetric Cryptography

Credential Access

T1111 Multi-Factor Authentication Interception

Execution

T1204.001 Malicious Link

Discovery

T1217 Browser Information Discovery; T1614 System Location Discovery

Initial Access

T1566.002 Spearphishing Link

Resource Development

T1583.001 Acquire Infrastructure

Impact

T1657 Financial Theft

initial-access

T1660 Phishing

Remediation

Immediate actions

  • Deploy the Suricata rule matching WebSocket URI content '/webSocket/QT/' co-occurring with the hardcoded trailing access-token '/khkjsahfjkwhakjlsdwdddddd88' — the highest-confidence JWR network fingerprint per Group-IB.
  • Alert on co-occurrence of all seven '/api/open/*' endpoints (addClick, getSyncSettings, addCvv, the_final_interface, updateStatus, getPendingInstruction, binLookup) on a single host.
  • Enable SMS-linked brand-abuse / Digital Risk Protection monitoring so newly registered lookalike domains are actioned within their sub-2-day median lifespan.

Workarounds

  • Instruct customers to never submit OTP, card, or banking credentials via SMS-delivered links; verify communications through official apps or websites only.
  • Deploy Advanced URL Filtering, Advanced DNS Security, and cloud-delivered security services to block disposable phishing domains at resolution time.

Longer-term hardening

  • Build detection on passive DNS + WHOIS + visual phishing-page clustering and registrar/nameserver graph analysis rather than lexical domain-name matching alone, per Unit 42's guidance on Dominet (HK)-registered infrastructure.
  • Track the JWR kit's HTML stage-prefix naming convention (a_ through z_) and the 'belong_to_template' / 'cfg.projectName' fields in captured exfil traffic to cluster and take down operator crews together.
  • Feed the published YARA rule (co-occurrence of /api/open/addCvv, /webSocket/QT, JWRCVV-, belong_to_template, khkjsahfjkwhakjlsdwdddddd88, i_payal_select_verify.html) into kit-hunting pipelines.

Affected and fixed versions

  • Multiple impersonated brands (regional banks, brokerages, telecom carriers, postal/logistics operators, government toll and fine agencies, PayPal) — SMS-delivered brand-impersonation phishing pages served by the JWR/Outsider PaaS kit (no software vulnerability — this is a phishing-kit campaign, not a CVE)

Timeline

Last updated · revalidated 1× · latest source

  • Unit 42 dates the broader Smishing Triad PhaaS campaign's start, with fraudulent toll-violation and package-misdelivery smishing distributed since this date.
  • Smishing Triad begins systematically targeting U.S. residents (Unit 42), predating the JWR/Outsider-specific tooling documented later.
  • Smishing Triad domain-registration pattern pivots from 'com-' to 'gov-' prefixes, coinciding with UAE/Emirates Post impersonation reporting from Resecurity.
  • Palo Alto Networks Unit 42 publishes 'The Smishing Deluge,' linking Smishing Triad to 194,000+ malicious domains across 121+ countries and over $1B in cumulative fraud proceeds.
  • Group-IB's observation window for the Outsider/ChenLun PaaS kit begins, ultimately recording 100,000+ phishing pages across 54+ countries through May 2026.
  • Google files a civil lawsuit against the Outsider group.
  • FBI cyber division announces a coordinated takedown partnership with Google and Lumen's Black Lotus Labs, dubbed 'Operation Ghost Hook,' targeting Outsider infrastructure.
  • Group-IB observes 700+ new Outsider phishing pages stood up within one month of the Google/FBI takedown action, showing affiliate resilience.
  • Group-IB releases kit-level detection artifacts alongside its JWR analysis: a YARA rule (4-of-7 strings), a Suricata rule on /webSocket/QT/ plus the hardcoded token, a Python decryption snippet, Jwr* localStorage keys, the JWRCVV victim ID format and the 2-second long-poll cadence.
  • Group-IB publishes 'Smish. Click. Drained,' detailing the JWR kit's Vue 2 architecture, AES-256-CTR WebSocket exfiltration, and 32-command real-time operator control plane used by the Outsider sub-cluster.

Update history

Sources cited

More from Outsider

Previous and next in phishing

Drafted by the Threadlinqs research pipeline from 9 sources; published ; last updated .

Threadlinqs Intelligence — Real-Time Threat Detection Platform

// threat_feed
$ sort --newest
Showing all threats

Live intelligence console

Threat level

Latest Threats