Back to Intelligence

k0t Edge Intrusion Campaign + M365 AiTM Token Replay: AdaptixC2, MeshCentral, FortiWeb/GlobalProtect CVE Detection Pack

SA
Security Arsenal Team
November 21, 2026
10 min read

Two TLP:WHITE OTX pulses converge on the same enterprise risk theme: identity compromise at the boundary, followed by durable access and monetizable account control. The first pulse, attributed to threat actor k0t, describes large-scale exploitation of network edge products — FortiGate, FortiWeb, and Palo Alto GlobalProtect — using CVE-2025-64446 against FortiWeb and CVE-2026-0257 against GlobalProtect, plus custom brute-force tooling. Reported outcomes include credential harvesting, persistent access, and initial-access-broker behavior, with more than 70,000 exposed credentials observed. Tooling includes AdaptixC2 and a MeshCentral endpoint at meshcentral.girydsfsggvb.com, indicating hands-on remote administration after exploitation.

The second pulse shifts from perimeter exploitation to cloud identity replay: an unknown actor used adversary-in-the-middle relays to capture Microsoft 365 sessions, then replayed stolen tokens from infrastructure attributed to German hosting provider 1337 Services GmbH. Between July and September 2026, Huntress reporting cited 79 compromised organizations and 497 successful sign-ins, consistent with business email compromise preparation or execution.

Collectively, these pulses describe an intrusion economy where edge CVE exploitation, credential brute forcing, AiTM session theft, and token replay can be chained without needing a noisy endpoint implant on every host. The likely attack chain is: expose or exploit edge services, harvest credentials and cookies/tokens, establish C2 or RMM persistence, replay cloud sessions, then monetize through BEC, data theft, or access resale. The OTX pulses leave MITRE ATT&CK fields blank, but behavior is consistent with T1190 Exploit Public-Facing Application, T1110 Brute Force, T1078 Valid Accounts, T1557 Adversary-in-the-Middle, T1528 Steal Application Access Token, T1105 Ingress Tool Transfer, and T1219 Remote Access Software.

Threat Actor / Malware Profile

k0t is reported as an initial access broker focused on network edge devices. Distribution is not spam-driven in the classic sense; access is created by scanning and exploiting exposed FortiGate, FortiWeb, and Palo Alto GlobalProtect surfaces, then escalating through credential theft, brute force, and persistence. The pulse references FortiWeb exploitation via CVE-2025-64446 and GlobalProtect exploitation via CVE-2026-0257, while also listing older Active Directory and FortiOS credential-related CVEs — CVE-2021-42278, CVE-2021-42287, and CVE-2019-6693 — that may reflect tooling knowledge, post-exploitation paths, or adjacent exposure checks rather than confirmed use in every intrusion.

AdaptixC2 is the named post-exploitation framework. Expect modular payload execution, configurable C2, process injection or spawning patterns, and operator-driven discovery. Because the pulse does not provide a beacon profile, treat C2 detection as behavior-led: unusual outbound HTTPS from edge-management hosts, rare user agents, periodic small-payload connections, and processes launched from temporary or web-accessible directories. Persistence is most plausibly achieved through valid accounts, edge device configuration changes, implanted web shells or services, and remote access tooling rather than a single registry run key.

MeshCentral activity is significant because legitimate RMM can provide persistence, interactive control, file transfer, and tunneling while blending into admin workflows. The indicator meshcentral.girydsfsggvb.com should be treated as high-signal but not sufficient alone: hunt for unauthorized meshagent, unexpected service installation, new local accounts, and connections from servers or edge-management jump hosts that normally do not run RMM.

The M365 AiTM/token replay actor is unknown and malware-family-free. Their payload is the stolen session itself. Distribution is likely phishing or adversary-in-the-middle relay pages that proxy Microsoft sign-in and capture tokens after MFA. Payload behavior is session reuse from foreign hosting infrastructure, mailbox rule creation, OAuth consent abuse, and BEC workflows. Persistence comes from refresh tokens, added MFA methods, mailbox forwarding, or malicious app consent. Anti-analysis is primarily identity-native: token replay bypasses many endpoint controls, uses valid sessions, and may leave minimal malware artifacts.

IOC Analysis

The strongest atomic indicators in these pulses are the MeshCentral hostname meshcentral.girydsfsggvb.com, the SHA256 hashes 39c4decc4881ec02a47cd01c0fd8c3030527b466124ad03c6f99cfaa0ba0b72e and d8a3fedb1a3cd1287ca47c6851a49e6052ab6116b43666e674765ca892d798ce, MD5 6f5d18157425aa9697913ae1310f588f, and CVEs tied to perimeter exposure: CVE-2025-64446, CVE-2026-0257, CVE-2021-42278, CVE-2021-42287, and CVE-2019-6693. The domain file.name appears malformed or placeholder-like and should be validated before blocking. No IPv4/IPv6 indicators were provided for 1337 Services GmbH; do not fabricate ranges. Use passive DNS, netflow, proxy, and EDR network telemetry to expand from confirmed infrastructure only.

SOC teams should operationalize indicators by type. Domains and hostnames go to DNS sinkhole, proxy block, EDR network connection search, and TLS/JA3 or certificate pivoting where available. Hashes go to EDR block lists, malware detonation, retro-hunt across process and file events, and email/secure-web-gateway detonation. CVEs go to exposure management: asset inventory for FortiGate, FortiWeb, GlobalProtect, domain controllers, and internet-facing VPN portals; validate patch level and configuration, not merely scanner presence. Token-replay intelligence maps better to identity analytics than static IOCs: impossible travel, token reuse across autonomous systems, sign-in from hosting providers, MFA method changes, mailbox rules, OAuth grants, and audit events after AiTM phishing. Tooling that decodes or enriches these indicators includes MDE/EDR retro-hunt, Sigma and KQL engines, Microsoft Sentinel watchlists, MISP/OTX subscriptions, firewall/VPN logs, Entra ID sign-in logs, and sandbox detonation for hash context.

Detection Engineering

YAML
---
title: k0t Edge Exploitation Followed by AdaptixC2 or RMM Staging
id: 9b0c9ad1-3b8d-4d1a-9f7a-6f7a4a7b1201
status: experimental
description: Detects suspicious process execution on or from edge-management systems consistent with exploitation of FortiWeb/GlobalProtect, web shell spawning, or AdaptixC2/RMM staging reported in OTX k0t pulse.
author: Security Arsenal
logsource:
  category: process_creation
  product: windows
detection:
  selection_parent:
    ParentImage|endswith:
      - '\w3wp.exe'
      - '\httpd.exe'
      - '\nginx.exe'
      - '\java.exe'
      - '\php-cgi.exe'
      - '\cmd.exe'
  selection_child:
    Image|endswith:
      - '\powershell.exe'
      - '\pwsh.exe'
      - '\rundll32.exe'
      - '\regsvr32.exe'
      - '\certutil.exe'
      - '\bitsadmin.exe'
      - '\curl.exe'
      - '\wget.exe'
      - '\meshagent.exe'
  filter_known_admin:
    CommandLine|contains:
      - 'healthcheck'
      - 'backup'
  condition: selection_parent and selection_child and not filter_known_admin
falsepositives:
  - Legitimate web application administration
  - Approved RMM deployment
level: high
tags:
  - attack.t1190
  - attack.t1105
  - attack.t1059
  - attack.t1219
---
title: Unauthorized MeshCentral Agent Install or Beacon
id: 7a15d32b-27c1-43e2-a96b-b55ce9ef7b22
status: experimental
description: Detects MeshCentral agent execution, service installation, or command lines referencing the OTX-reported C2/RMM hostname meshcentral.girydsfsggvb.com.
author: Security Arsenal
logsource:
  category: process_creation
  product: windows
detection:
  selection_hash:
    Hashes|contains:
      - 'SHA256=39C4DECC4881EC02A47CD01C0FD8C3030527B466124AD03C6F99CFAA0BA0B72E'
      - 'MD5=6F5D18157425AA9697913AE1310F588F'
  selection_mesh:
    Image|endswith:
      - '\meshagent.exe'
      - '\MeshCentral.exe'
    CommandLine|contains:
      - 'meshcentral.girydsfsggvb.com'
      - 'meshagent'
      - 'agent install'
  selection_service:
    Image|endswith: '\sc.exe'
    CommandLine|contains|all:
      - 'create'
      - 'mesh'
  condition: selection_hash or selection_mesh or selection_service
falsepositives:
  - Authorized MeshCentral or other RMM used by IT
level: critical
tags:
  - attack.t1219
  - attack.t1543.003
  - attack.t1105
---
title: Credential Access Artifacts After Edge or Token Intrusion
id: 2f6c8c34-e2a7-40db-a1c9-8f76a927db10
status: experimental
description: Detects suspicious access to LSASS, credential stores, browser login data, or token/cache locations that may follow k0t credential harvesting or M365 AiTM token theft workflows.
author: Security Arsenal
logsource:
  category: process_access
  product: windows
detection:
  selection_lsass:
    TargetImage|endswith: '\lsass.exe'
    GrantedAccess|contains:
      - '0x1010'
      - '0x1410'
      - '0x143a'
      - '0x1fffff'
  selection_source_suspicious:
    SourceImage|endswith:
      - '\powershell.exe'
      - '\rundll32.exe'
      - '\regsvr32.exe'
      - '\procdump.exe'
      - '\comsvcs.dll'
      - '\meshagent.exe'
      - '\adaptix.exe'
  filter_tools:
    SourceImage|startswith:
      - 'C:\Program Files\Microsoft Defender'
      - 'C:\Windows\System32\svchost.exe'
  condition: selection_lsass and selection_source_suspicious and not filter_tools
falsepositives:
  - EDR, backup, and approved admin tools accessing LSASS
level: high
tags:
  - attack.t1003.001
  - attack.t1552
  - attack.t1528
KQL — Microsoft Sentinel / Defender
let Lookback = 14d;
let BadHost = "meshcentral.girydsfsggvb.com";
let HashSha256 = dynamic(["39c4decc4881ec02a47cd01c0fd8c3030527b466124ad03c6f99cfaa0ba0b72e","d8a3fedb1a3cd1287ca47c6851a49e6052ab6116b43666e674765ca892d798ce"]);
let HashMd5 = dynamic(["6f5d18157425aa9697913ae1310f588f"]);
let EdgeTerms = dynamic(["fortigate","fortiweb","globalprotect","panw","paloalto","vpn","sslvpn"]);
DeviceNetworkEvents
| where TimeGenerated >= ago(Lookback)
| where RemoteUrl has BadHost or RemoteUrl contains "meshcentral" or RemoteIP has_any (dynamic([]))
| project TimeGenerated, DeviceName, InitiatingProcessFileName, InitiatingProcessCommandLine, RemoteUrl, RemoteIP, RemotePort, ActionType
| extend Hunt = "MeshCentral/RMM network"
| union (
    DeviceProcessEvents
    | where TimeGenerated >= ago(Lookback)
    | where SHA256 has_any (HashSha256) or MD5 has_any (HashMd5)
       or FileName has_any (dynamic(["meshagent.exe","MeshCentral.exe","adaptix.exe","beacon.exe"]))
       or ProcessCommandLine has_any (dynamic([BadHost,"meshagent","adaptix","fortiweb","globalprotect","cve-2025-64446","cve-2026-0257"]))
    | project TimeGenerated, DeviceName, AccountName, FileName, ProcessCommandLine, SHA256, MD5, InitiatingProcessFileName, InitiatingProcessCommandLine
    | extend Hunt = "Hash/RMM/process"
)
| union (
    SecurityEvent
    | where TimeGenerated >= ago(Lookback)
    | where EventID in (4624,4625,4648,4740,4768,4769,4771,4776)
    | where (EventID == 4625 and (Account has_any (EdgeTerms) or WorkstationName has_any (EdgeTerms) or IpAddress has_any (EdgeTerms)))
       or (EventID == 4624 and LogonType in (3,10) and (Account has_any (EdgeTerms) or WorkstationName has_any (EdgeTerms)))
    | project TimeGenerated, Computer, Account, LogonType, AuthenticationPackageName, IpAddress, WorkstationName, EventID, Status, SubStatus
    | extend Hunt = "Edge auth anomaly"
)
| sort by TimeGenerated desc
PowerShell
# Read-only hunt for k0t/AdaptixC2/MeshCentral artifacts and M365 token-theft fallout on Windows endpoints.
$ErrorActionPreference = 'SilentlyContinue'
$BadHost = 'meshcentral.girydsfsggvb.com'
$Sha256 = @('39c4decc4881ec02a47cd01c0fd8c3030527b466124ad03c6f99cfaa0ba0b72e','d8a3fedb1a3cd1287ca47c6851a49e6052ab6116b43666e674765ca892d798ce')
$Md5 = @('6f5d18157425aa9697913ae1310f588f')
$Report = [ordered]@{ Hostname=$env:COMPUTERNAME; Time=(Get-Date).ToUniversalTime(); Hits=@() }
function Add-Hit($Type,$Detail){ $Report.Hits += [pscustomobject]@{ Type=$Type; Detail=$Detail } }

# 1) DNS and live connections to reported RMM/C2 hostname.
try {
  Resolve-DnsName $BadHost | ForEach-Object { Add-Hit 'DNS' ($BadHost + ' -> ' + $_.IPAddress) }
} catch {}
Get-NetTCPConnection -State Established | ForEach-Object {
  $p = Get-Process -Id $_.OwningProcess
  if ($_.RemoteAddress -and ($p.ProcessName -match 'mesh|adaptix|beacon|powershell|rundll32|regsvr32' -or $_.RemotePort -in 443,80,8080,8443)) {
    Add-Hit 'NetTCP' (($p.ProcessName) + ' pid=' + $_.OwningProcess + ' remote=' + $_.RemoteAddress + ':' + $_.RemotePort)
  }
}

# 2) Unauthorized MeshCentral/RMM processes, services, run keys, and scheduled tasks.
Get-Process | Where-Object { $_.ProcessName -match 'meshagent|MeshCentral|adaptix|beacon' } | ForEach-Object { Add-Hit 'Process' ($_.ProcessName + ' pid=' + $_.Id + ' path=' + $_.Path) }
Get-CimInstance Win32_Service | Where-Object { $_.Name -match 'mesh|adaptix|beacon' -or $_.PathName -match 'meshagent|adaptix|beacon' } | ForEach-Object { Add-Hit 'Service' ($_.Name + ' state=' + $_.State + ' path=' + $_.PathName) }
$RunKeys = @('HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Run','HKCU:\SOFTWARE\Microsoft\Windows\CurrentVersion\Run','HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\RunOnce','HKCU:\SOFTWARE\Microsoft\Windows\CurrentVersion\RunOnce')
foreach ($k in $RunKeys) { Get-ItemProperty $k | ForEach-Object { $_.PSObject.Properties | Where-Object { $_.Value -match 'meshagent|adaptix|beacon|powershell -enc|rundll32' } | ForEach-Object { Add-Hit 'RunKey' ($k + ' ' + $_.Name + '=' + $_.Value) } } }
Get-ScheduledTask | Where-Object { $_.TaskName -match 'mesh|adaptix|beacon' -or ($_.Actions | Out-String) -match 'meshagent|adaptix|beacon' } | ForEach-Object { Add-Hit 'ScheduledTask' ($_.TaskName + ' actions=' + ($_.Actions | Out-String)) }

# 3) Hash sweep in high-risk staging paths only; do not delete from this script.
$Paths = @($env:TEMP,$env:ProgramData,'C:\Windows\Temp','C:\Users\Public')
foreach ($base in $Paths) {
  Get-ChildItem $base -Recurse -File -Include *.exe,*.dll,*.ps1,*.bat,*.js,*.jar -ErrorAction SilentlyContinue | ForEach-Object {
    $h = Get-FileHash $_.FullName -Algorithm SHA256
    $m = Get-FileHash $_.FullName -Algorithm MD5
    if ($Sha256 -contains $h.Hash.ToLower() -or $Md5 -contains $m.Hash.ToLower()) { Add-Hit 'HashHit' ($_.FullName + ' SHA256=' + $h.Hash + ' MD5=' + $m.Hash) }
  }
}

# 4) Token/credential theft surface: unusual browser credential files copied to temp, new local admins, recent MFA-like artifacts.
Get-ChildItem ($env:TEMP,'C:\Windows\Temp') -Recurse -File -Include *Login*Data*,*Cookies*,*Web*Data*,*.ost,*.pst -ErrorAction SilentlyContinue | ForEach-Object { Add-Hit 'CredentialStoreCopy' $_.FullName }
Get-LocalGroupMember -Group 'Administrators' | ForEach-Object { Add-Hit 'LocalAdmin' $_.Name }
[pscustomobject]$Report | ConvertTo-Json -Depth 6

Response Priorities

Immediate: block and alert on meshcentral.girydsfsggvb.com at DNS, proxy, and EDR network controls; add the two SHA256 values and MD5 to EDR prevention and retro-hunt; isolate any host with unauthorized meshagent, AdaptixC2 artifacts, or LSASS access from non-admin tools; verify FortiGate, FortiWeb, and GlobalProtect exposure and emergency-patch or virtual-patch CVE-2025-64446 and CVE-2026-0257; temporarily restrict edge management interfaces to known admin networks.

24h: because both pulses are identity-centric, force verification of credentials and sessions, not just endpoints. Reset credentials harvested or used from edge devices, revoke Entra ID refresh tokens for affected users, require phishing-resistant MFA re-registration where AiTM is possible, review MFA method changes, mailbox rules, forwarding, OAuth consents, and sign-ins from hosting providers. Search VPN, firewall, WAF, GlobalProtect, FortiWeb, and Entra logs for brute force bursts, impossible travel, token reuse, and post-sign-in mailbox changes. Validate whether any of the 70,000 exposed credentials overlap your directory through a controlled credential-breach workflow.

1 week: harden architecture around the observed vector: remove edge device management from the public internet, enforce FIDO2/passkeys or number-matching plus token protection where supported, enable Continuous Access Evaluation and Conditional Access token binding, deploy browser isolation for high-risk sign-ins, and segment edge appliances from domain controllers and SCCM/RMM planes. Establish detections for valid-account anomalies after edge exploitation, alert on new RMM services anywhere outside approved software, and run attack surface monitoring for Fortinet/Palo Alto banners, exposed portals, certificate changes, and lookalike AiTM domains.

Related Resources

Security Arsenal Incident Response Managed SOC & MDR Services AlertMonitor Threat Detection From The Dark Side Intel Hub

Is your security operations ready?

Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.