Skip to navigation

Threat reportPhishingTL-2026-1245

Compromised RD Session Host Used to Stage Boots-Themed Phishing Campaign via Gammadyne Mailer

mediumACTIVE

A Romanian-linked operator brute-forced and then logged into an internet-exposed, MFA-less Microsoft RDWeb/RD Gateway portal at a small business using stolen credentials, staged the legitimate bulk-mail tool Gammadyne Mailer on the RD Session Host, and used it to blast phishing emails impersonating UK pharmacy chain Boots to a target list of 8,894,920 addresses.

Key facts

IdentifiersTL-2026-1245
Also known asBoots Phishing Campaign, Terminal Server Phishing Stager, The Devil, Eight Million Emails, and a Whole Lot of Milk
SeverityMEDIUM
Affected productsMicrosoft Remote Desktop Web Access (RDWeb) / RD Gateway / RD Session Host
Weaknesses (CWE)CWE-287, CWE-307, CWE-521, CWE-778
MalwareGammadyne Mailer, Turbo-Mailer
Toolsdracii.mmp (Gammadyne project file)
MITRE ATT&CK techniquesT1005 Data from Local System, T1018 Remote System Discovery, T1021.001 Remote Desktop Protocol, T1036 Masquerading, T1071.003 Mail Protocols, T1078.002 Domain Accounts, T1110.001 Password Guessing, T1114 Email Collection, T1133 External Remote Services, T1187 Forced Authentication, T1204.002 Malicious File, T1219 Remote Access Tools and 7 more in the attack flow
MotivationFINANCIAL
Target sectorssmall business, retail, government administration, managed service providers, consumer general public
Target regionsunited kingdom, bolivia, romania, Global
Detection rules9 (1 shown free)
Sources6 cited sources

Detection coverage

As of 2026-06-15, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-1245 across Splunk SPL, Microsoft KQL and Sigma, covering 34 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: RDWeb Forms-Auth High-Volume Login POST Burst (Brute Force)

Detects a high-volume burst of forms-based authentication POST requests against an internet-facing RDWeb/RD Gateway portal from a small set of source IPs — the pattern observed from 87.251.64.134 and the 88.210.63.0/24 range, and consistent with generic Go-http-client-style credential-guessing bots hitting /RDWeb/Pages/ before a valid-account login.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1110.001, T1133
  • Data sources: iis, windows_security
  • Severity: high
  • Confidence: high

Splunk SPL

index=iis sourcetype="ms:iis:auto" cs_method=POST cs_uri_stem="*/RDWeb/Pages/*"
| bin _time span=5m
| stats count AS post_count dc(cs_uri_stem) AS path_variety values(cs_user_agent) AS user_agents by c_ip, _time
| where post_count > 100
| eval risk=if(match(user_agents, "(?i)go-http-client"), "critical", "high")
| table _time, c_ip, post_count, path_variety, user_agents, risk
| sort -post_count

Microsoft KQL

let threshold = 100;
W3CIISLog
| where TimeGenerated > ago(24h)
| where csMethod == "POST" and csUriStem has "/RDWeb/Pages/"
| summarize PostCount = count(), UserAgents = make_set(csUserAgent) by cIP = cIp, bin(TimeGenerated, 5m)
| where PostCount > threshold
| project TimeGenerated, cIP, PostCount, UserAgents
| order by PostCount desc

Sigma

title: RDWeb High-Volume Forms-Auth Login POST Burst
id: 7c1e2a4a-1b3c-4e2f-9a6d-1f2b3c4d5e01
status: experimental
description: Detects a burst of RDWeb/RD Gateway forms-authentication login POST requests from a single source IP consistent with credential brute-forcing against an internet-exposed portal.
references:
    - https://www.huntress.com/blog/terminal-server-phishing-stager-exposed
author: Threadlinqs Team
date: 2026-07-11
tags:
    - attack.credential-access
    - attack.t1110.001
    - attack.initial-access
    - attack.t1133
logsource:
    product: windows
    service: iis
detection:
    selection:
        cs-method: 'POST'
        cs-uri-stem|contains: '/RDWeb/Pages/'
    timeframe: 5m
    condition: selection | count(c-ip) by c-ip > 100
falsepositives:
    - Legitimate high-volume RDWeb usage during mass password-reset or onboarding events (rare)
level: high

Known false positives:

  • Legitimate bulk password-reset campaigns or vulnerability-scanner authorized testing against the same portal

Tuning notes: Tune the 100-POST/5-minute threshold to the organization's normal RDWeb login volume; exclude known vulnerability-scanner source IPs; correlate with cs-user-agent for Go-http-client or other non-browser clients to raise confidence.

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)
  • RDP Session Reconnection to Existing Disconnected Session Without Preceding 4624 · unlock with TL Creds
  • Signed Bulk-Mail Utility Execution on Non-Mail-Relay Terminal Server (Living-off-the-Land Mailer Abuse) · unlock with TL Creds
  • Themed Mailer-Kit Staging Directory and Recipient-List File Creation · unlock with TL Creds
  • Direct-to-MX High-Concurrency Outbound TCP/25 Burst From Non-Mail-Relay Host · unlock with TL Creds
  • Compromised Third-Party Web Service Hosting Phishing Credential-Harvest Kit (Compromise Infrastructure: Web Services) · unlock with TL Creds
  • Merge-Field Phishing Subject Line and Cosmetic Fake-PGP Block Consistent With Operator Kit · unlock with TL Creds
  • Lone Successful RDWeb Login After Extended Failure Streak (Credential Stuffing Hit) · unlock with TL Creds
  • Human-Operator RDWeb Access Pattern (.rdp Retrieval + Browser-Realistic UA + Domain-Qualified Username) · unlock with TL Creds

Indicators of compromise

As of 2026-06-15, Threadlinqs Intelligence tracks 34 indicators of compromise for TL-2026-1245.

Sample indicators (3 of 34)

  • IP address: 212.93.152[.]37
  • Domain: ipelc[.]gob[.]bo
  • File hash: dc8a31a104df44b76a05c3e6d4fbe85be243cc6db0ff25e298e25cc469715046

Defanged, TLP:CLEAR. Full set of 34 indicators: unlock with TL Creds.

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

Technical analysis

The credential-harvesting payload was hosted on a compromised Bolivian government institute website; Huntress detected the mail-sending burst within minutes, isolated all 25 endpoints on the host, and blocked 29,954 further outbound SMTP connections.

Between 2026-05-13 and 2026-05-16, an internet-facing Microsoft RD Session Host with an exposed RDWeb portal (IIS front end, RD Gateway role enabled, forms-based authentication, no MFA) at a small business was targeted by at least two distinct attacker populations. Over the four days preceding the incident the portal received 657,521 requests from 8,673 distinct IPs, including more than 206,000 login POST attempts (successful logins are distinguishable in IIS logs by an HTTP 302 redirect vs. HTTP 200 for failures). High-volume brute-force bots included 87.251.64.134 (~48,000 requests, no success) and the 88.210.63.0/24 range (~84,000 requests across roughly 10 hosts, no success), both using bare usernames, lowercase portal paths, and generic Go-http-client/2.0 user agents. A separate bot, 216.152.151.168, achieved a single successful login after 127 failed POST attempts on 2026-05-15 21:51 UTC but never pulled a .rdp file or opened a session. A human operator using a browser-realistic client (mixed-case portal paths, domain-qualified usernames, spoofed Mozilla/5.0 user-agent variants, and actual .rdp file retrieval) accessed the portal from Romanian IPs 80.94.95.37 (Timisoara-area hosting, already flagged in threat feeds, first seen 2026-05-13 19:21 UTC) and 212.93.152.37 (Romanian residential fixed-line, first seen 2026-05-15 02:12 UTC). The domain account used validated successfully four separate times across the observed activity (two by the hands-on operator, one by the brute-forcer, one apparent credential-dump hit), consistent with stolen/leaked credentials rather than a software exploit.

The operator's RDP client workstation, tracked in Huntress telemetry as DESKTOP-[REDACTED], had previously been observed across multiple unrelated partner incidents, indicating a shared or rotated attacker toolkit/VM rather than a one-off machine. On 2026-05-16 03:36 UTC the operator (212.93.152.37) reconnected to an existing, disconnected RDP session (session 16) rather than establishing a fresh logon — visible in IIS Event 1315 (forms-auth ticket re-authentication, 03:36:00), RD Gateway Events 312/200/300/302 (tunnel establishment and gateway authentication, 03:36:14), TS-RemoteConnectionManager Event 1149 (RDP authentication success, 03:36:18), and TS-LocalSessionManager Event 25 (session reconnection, 03:36:28). Windows Security auditing on the host was minimal — only two 4624 logon events were recorded in total despite the extensive brute-force and session activity — making the IIS/RDWeb, RD Gateway, and Terminal Services channels the primary source of forensic evidence rather than the Security log.

On the compromised host the operator staged a folder named 'dam pe uk puterniiicccc' (Romanian slang for 'we hit the UK hard') on the desktop, containing the legitimate commercial bulk-mail application Gammadyne Mailer v11.x (gm.exe, digitally signed, zero antivirus detections), a project file named dracii.mmp ('the devils' in Romanian, created 2025-07-18 — roughly ten months before the incident — and last modified 2026-05-16 03:28 UTC) holding sender-spoofing configuration, phishing lure/subject-line templates, and payload URLs, plus six recipient-list text files ('milk (1).txt' through 'milk (6).txt') totaling 8,894,920 email addresses (approximately 1M, 1.33M, 685K, 3.88M, 1M, and 1M addresses respectively). The dracii.mmp configuration also referenced a UNC path to a previously compromised RDS domain (\\[REDACTED].local\RDS\RDSRedirections\...\Crack\gm.log), further indicating the kit had been carried across multiple prior victim terminal servers rather than freshly built for this target. Huntress's broader visibility into the operator's toolkit (via related incidents) showed additional, unused recipient-list themes named for other UK-centric lures — including HMRC (UK tax authority) and Solana cryptocurrency holders — and files with names such as 'fara gmail' (Romanian for 'without Gmail'), indicating an operational rotation of phishing themes rather than a single-campaign toolkit. The mailer was configured for direct-to-MX delivery (performing its own MX lookups and connecting straight to recipient mail servers rather than routing through the victim's own outbound mail relay, and using no DKIM/DomainKeys signing or smart host), with 666 concurrent send threads in 50-message batches, and an operator seed/test address of lacafea77@outlook.com embedded in the Gammadyne configuration. Direct-to-MX delivery both evades mail-provider account throttling and protects the attacker's own sending reputation by burning the victim's IP address instead of a tenant mail domain.

The phishing lure impersonated Boots UK ('boots' <hello@boots.com>), using a merge-field-personalized subject line ('[[-Now-]] Share Your Feedback & Receive a Free Gift from Boots (@bootsUK)... Customer satisfaction survey E-mail:[[-Email-]] NO:[[random_digits(7)]]-[[random_digits(5)]]') driving recipients through a cloned Boots storefront landing page, a throwaway 'customer satisfaction survey' modal, a 'Secure Checkout' form harvesting full name, email, date of birth, phone number, and home address, and a final payment step harvesting payment-card data — all behind a bogus £0.00 order flow and a fake 'Skinny Tan' sponsor tie-in. Seven additional, unused subject-line/body-copy variants were found stored in the kit but never deployed in this run. A cosmetic PGP-signature block (styled HTML text mimicking BEGIN/END PGP SIGNED MESSAGE, Hash: SHA512, and BEGIN/END PGP SIGNATURE, with no actual cryptographic capability) was embedded purely to add visual legitimacy to the phishing page.

Huntress's MDR agent, deployed on the host on 2026-05-15 18:48 UTC (roughly 33 hours before the burst), raised its first critical detection at 2026-05-16 03:38:53 UTC as gm.exe began firing outbound SMTP connections. Over a roughly 104-second window (03:47:03–03:48:47 UTC) the mailer attempted connections to 1,641 distinct MX servers (Google, Microsoft, Yahoo, and regional providers) on TCP/25. The Huntress SOC mass-isolated all 25 endpoints on the customer network via a default-deny filter, flagged the compromised domain account for partner-side reset, severed the active RDP session, blocked further reconnection attempts from 212.93.152.37, and — per Windows Filtering Platform Event 5157 (logging the specific process path \device\harddiskvolume4\users\[user]\desktop\dam pe uk puterniiicccc\gm.exe) — blocked 29,954 additional outbound TCP/25 connection attempts post-isolation. Because delivery was direct-to-MX, the victim's own Microsoft 365/Google mail tenant was never used as a relay, limiting reputational blast radius to the victim's outbound IP rather than its mail domain; the true total of messages successfully sent before isolation cannot be quantified from the available logs, since WFP Event 5157 only captures connections blocked after isolation began. Huntress reported the compromised Bolivian government domain (ipelc.gob.bo, operated by the Instituto Plurinacional de Estudio de Lenguas y Culturas) and its /boots_store/ phishing kit to Centro de Gestión de Incidentes Informáticos (CGII), Bolivia's national CSIRT, coordinated through AGETIC, the country's state ICT agency.

The underlying root cause was a purely configuration-level exposure — an internet-facing RDWeb/RD Gateway portal with no MFA — rather than a software vulnerability, and Gammadyne Mailer itself is legitimate commercial software whose weaponization here is consistent with prior reported abuse of similar bulk-mailers (e.g., Gammadyne Mailer and Turbo-Mailer) by BEC- and phishing-focused actors, including the Nigeria-linked 'TMT' gang, for large-scale direct-to-MX email delivery from compromised infrastructure.

MITRE ATT&CK techniques by tactic:

Collection

T1005 Data from Local System; T1114 Email Collection

Discovery

T1018 Remote System Discovery

lateral-movement

T1021.001 Remote Desktop Protocol

Defense Evasion

T1036 Masquerading

Command and Control

T1071.003 Mail Protocols; T1219 Remote Access Tools

Initial Access

T1078.002 Domain Accounts; T1133 External Remote Services; T1566.002 Spearphishing Link

Credential Access

T1110.001 Password Guessing; T1187 Forced Authentication

Execution

T1204.002 Malicious File

Lateral Movement

T1570 Lateral Tool Transfer

Resource Development

T1583.003 Virtual Private Server; T1584.001 Domains; T1584.006 Web Services

Reconnaissance

T1589.002 Email Addresses

reconnaissance

T1598.003 Spearphishing Link

Remediation

Immediate actions

  • Isolate the compromised RD Session Host and any endpoints sharing the compromised domain account
  • Disable/reset the compromised domain account and force credential rotation for all accounts with RDWeb access
  • Block outbound TCP/25 from non-mail-relay hosts, and block the identified attacker IPs (212.93.152.37, 80.94.95.37, 216.152.151.168, 87.251.64.134, 88.210.63.0/24) at the perimeter
  • Remove Gammadyne Mailer (gm.exe), dracii.mmp, and the 'milk (1-6).txt' recipient lists from the staging directory and preserve for forensics
  • Report the compromised payload-hosting domain (ipelc.gob.bo) to the relevant national CSIRT for takedown
  • Search environment-wide for the reused RDP client workstation identifier and the UNC path pattern (...\RDS\RDSRedirections\...\Crack\gm.log) as a pivot for other compromised RDS hosts tied to the same operator

Workarounds

  • If RDWeb cannot be immediately removed from the internet, restrict access via IP allowlisting and require client certificate authentication until MFA/VPN is deployed

Longer-term hardening

  • Remove direct internet exposure of RDWeb/RD Gateway; place behind a VPN or ZTNA/SASE gateway
  • Enforce MFA on all remote-access portals, including RDWeb and RD Gateway
  • Enable and centralize verbose Windows Security auditing (4624/4625) and Terminal Services event logging, forwarded to a SIEM
  • Deploy account lockout / adaptive throttling on RDWeb authentication endpoints to blunt brute-force campaigns
  • Monitor for anomalous high-volume outbound SMTP (TCP/25) from hosts that are not designated mail relays
  • Baseline and alert on execution of commercial bulk-mail utilities (Gammadyne Mailer, Turbo-Mailer, and similar) on non-mail-server endpoints

Affected and fixed versions

  • Microsoft — Remote Desktop Web Access (RDWeb) / RD Gateway / RD Session Host
    Vulnerable versions: Any RDWeb/RD Gateway deployment exposed directly to the internet without MFA
    Fixed in: N/A — requires configuration hardening (MFA, VPN/ZTNA placement), not a software patch

Timeline

  • Gammadyne Mailer project file dracii.mmp is created (per its internal Gammadyne timestamp), the earliest evidence of campaign tooling preparation, roughly ten months before the incident.
  • Human operator logs into the exposed RDWeb portal and pulls a published .rdp file from source IP locked-ioc-19 (Romania, Timisoara hosting) at 19:21 UTC.
  • A separate brute-force bot at locked-ioc-14 achieves one successful login after 127 failed attempts at 21:51 UTC but never pulls a .rdp file or opens a session.
  • Huntress MDR agent is deployed to the customer's environment at 18:48 UTC, roughly 33 hours before the malicious mail-sending burst.
  • Operator logs into RDWeb portal and pulls a .rdp file from locked-ioc-16 (Romania, residential) at 02:12 UTC, the IP used through the rest of the intrusion.
  • Gammadyne Mailer fires a 104-second burst of connection attempts to 1,641 distinct mail servers between 03:47:03 and 03:48:47 UTC; Huntress SOC isolates all 25 endpoints and blocks the activity in progress, logging 29,954 blocked outbound TCP/25 attempts via Windows Filtering Platform Event 5157.
  • Huntress raises its first critical detection at 03:38:53 UTC as gm.exe begins outbound SMTP activity.
  • Operator (locked-ioc-16) reconnects to existing RDP session 16 at 03:36 UTC via IIS Event 1315, RD Gateway Events 312/200/300/302, TS-RemoteConnectionManager Event 1149, and TS-LocalSessionManager Event 25 — not a new logon.
  • Huntress publishes its blog post 'The Devil, Eight Million Emails, and a Whole Lot of Milk', disclosing the incident and reporting the compromised Bolivian government domain to CGII, coordinated via AGETIC.

Sources cited

Previous and next in phishing

Drafted by the Threadlinqs research pipeline from 6 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