
Operation Master: Deconstructing a Multi-Tiered Intrusion and Monetization Pipeline Threat Campaign
Indicators of Compromise
APT Groups1
Overview SOCRadar Threat Research Unit (STRU) has uncovered an end-to-end cybercrime operation dubbed Operation Master, in which a threat actor compromised enterprise networks, stole customer and billing databases, offered portions of the data for sale on underground forums, and later repurposed those exact assets to power an automated, multi-tenant invoice fraud platform. Operating across multiple countries, the threat actor compromised critical network infrastructure - including VPN gateways via a GlobalProtect authentication bypass (CVE-2026-0257) - while concurrently executing advanced web application exploits, deploying the AdaptixC2 framework, and leveraging an AI-assisted development workflow. Key Findings The threat actor exploited CVE-2026-0257 (GlobalProtect authentication bypass) to establish unauthorized network access across 7 gateways in four countries. Reconnaissance operations included 277.5 million address scans across 22 log files to curate a target list of 81 high-value organizations. Data theft was confirmed across 9+ database instances, leveraging xp_cmdshell to pivot from SQL injection to remote command execution. Offline credential artifacts, including SAM, SYSTEM, and SECURITY registry hives alongside an Active Directory database, were successfully exfiltrated. The threat actor utilized domainless phishing tactics, such as M365 OAuth device-code phishing and phone-guided vishing, bypassing traditional email gateways. Deployed AdaptixC2 as Command&Control Framework, establishing active control over 2 or more Windows server identities. Data exfiltration relied on dual redundant vectors, combining continuous DNS subdomain tunneling with automated rclone cloud synchronization. Operated an invoice-fraud engine generating over 2.4 million email/SMS messages and 622,666 personalized links before going offline in mid-September. Collected funds via PIX using a serverless proxy structure; receiving-side key material remains unrecoverable from this dataset. Tracked a single operator email address “cyberkill2025[@]gmail.com” across five distinct operational contexts and linked it to forum persona "masterblack" via leaked database records.
Campaign Guidance
Remediation, mitigation, notes, history and related intelligence
Technique ID | Technique Name | Detection Strategy |
|---|---|---|
Exploit Public-Facing Application | MITRE Detection Strategy DET0080 / AN0219: Adversary sends crafted input to an internet-facing app; the chain runs abnormal request patterns to a public endpoint, then elevated 4xx or 5xx or unusual methods, then a server process spawning a shell or LOLbins. [Campaign: the GlobalProtect forge chain and SQLi campaigns hit VPN and web endpoints; watch for the auth endpoint hit followed by anomalous session issuance.] | |
External Remote Services | MITRE Detection Strategy DET0354 / AN1004: Unusual external remote access followed by a successful session from uncommon geolocations or outside business hours, then internal lateral movement or exfiltration. [Campaign: recovered output shows working VPN sessions with assigned addresses and routes on seven gateways; alert on GlobalProtect sessions with forged-cookie fingerprints.] | |
Command and Scripting Interpreter: Windows Command Shell | MITRE Detection Strategy DET0202 / AN0578: Detects abuse of cmd.exe or shell invocation chains, focusing on parent-child relationships where cmd is launched from unusual parents and chained with discovery or credential access. [Campaign: sqlservr.exe spawning cmd.exe or powershell.exe via xp_cmdshell is the operation's pivot from SQL injection to host execution.] | |
OS Credential Dumping: Security Account Manager | MITRE Detection Strategy DET0085 / AN0235: A SYSTEM-level process accesses registry keys to dump the SAM hive, often with sequential access to HKLM\SAM and HKLM\SYSTEM and creation of .save or .dmp files. [Campaign: the operator extracted base64-encoded SAM, SYSTEM and SECURITY hives for offline cracking under a SQL service identity.] | |
OS Credential Dumping: NTDS | MITRE Detection Strategy DET0586 / AN1611: Detects NTDS.dit dumping by monitoring shadow copy creation, suspicious access to ntds.dit, and use of tooling such as ntdsutil.exe. [Campaign: a valid Active Directory database was recovered, and Evil-WinRM history shows attempts to copy the AD database from a tunneled domain controller.] | |
Exfiltration Over Unencrypted Non-C2 Protocol | MITRE Detection Strategy DET0149 / AN0423: Detects data access or staging followed by outbound flows over unencrypted protocols initiated by unexpected processes or to rare destinations. [Campaign: hunt for high-entropy DNS queries to unknown resolvers; a single host emitted 470,268 marker-prefixed subdomain queries to the operator's sink.] | |
Exfiltration to Cloud Storage | MITRE Detection Strategy DET0570 / AN1571: Unusual processes accessing large local files then initiating HTTPS to cloud-storage domains; correlate file reads in sensitive directories with uploads. [Campaign: a single-use rclone script filters roughly 140 sensitive file types and syncs to cloud storage with eight-stream transfers.] | |
Application Layer Protocol: Web Protocols | MITRE Detection Strategy DET0027 / AN0075: Detects unexpected or high-volume HTTP or WebSocket communication from suspicious processes using uncommon user agents or mimicking browser traffic to unusual domains. [Campaign: the AdaptixC2 HTTP listener on port 8443 and WebSocket API on 4321 at 91.92.241[.]187 carry gopher beacon traffic.] | |
Non-Application Layer Protocol | MITRE Detection Strategy DET0457 / AN1254: Anomalous use of non-web protocols by processes that do not normally generate such traffic, especially where flows bypass proxy infrastructure. [Campaign: the AdaptixC2 GOPHER_4444 TCP listener on port 4444 is raw-socket C2 outside normal web egress.] | |
Server Software Component: Web Shell | MITRE Detection Strategy DET0394 / AN1108: Unexpected file creation in web directories followed by web server processes such as w3wp.exe spawning command shells or script interpreters. [Campaign: minimal JSP command-execution webshells and a Metasploit WAR payload were staged against Tomcat hosts reached after VPN pivoting.] | |
Scheduled Task/Job: Scheduled Task | MITRE Detection Strategy DET0441 / AN1221: Detects creation or modification of scheduled tasks via Task Scheduler, WMI, PowerShell or API, including hidden or anomalous tasks under suspicious contexts. [Campaign: persistence was attempted through scheduled tasks with plausible service-like names, though durable persistence is not confirmed.] | |
Proxy: Multi-hop Proxy | MITRE Detection Strategy DET0359 / AN1020: Suspicious processes launch sustained encrypted outbound traffic to anonymity infrastructure and may relay to internal systems via reverse proxying or socket forwarding. [Campaign: scan and exploit origin is masked behind proxy hops, and a Ligolo-style reverse tunnel carries internal routing across two /24 ranges.] | |
Financial Theft | MITRE Detection Strategy DET0495 / AN1361: Monitor for mass emailing of payment instructions, sudden mailbox rule changes for financial staff, and anomalous access to financial or ERP systems. [Campaign: the master-panel dispatches fake-invoice email, SMS and WhatsApp at scale; hunt lookalike domains and the invoice URL schema carrying ?ref=, &venc= and &v0=.] | |
Access Token Manipulation: Token Impersonation/Theft | MITRE Detection Strategy DET0482 / AN1324: Detects token duplication and impersonation by correlating suspicious command lines with API calls such as DuplicateTokenEx or ImpersonateLoggedOnUser. [Campaign: GodPotato is staged as C# source and a compiled .NET binary for token-impersonation local privilege escalation from service contexts.] | |
Obfuscated Files or Information | MITRE Detection Strategy DET0378 / AN1064: Correlates script execution or suspicious parent processes with creation of encoded or encrypted file formats and abnormal command-line or PowerShell obfuscation. [Campaign: shellcode is XOR-wrapped with keys 0x37 and 0x41 and the FUD loader adds AES-128-CBC with hardcoded keys before execution.] |