LSN โ€” Local Sovereign Network

๐Ÿ”ด ACTIVE โ€” Krane/mdrfckr Multi-Actor Crypto-Mining Campaign ยท 32 IPs ยท 16 Countries ยท 20 ASNs ยท 55+ Days ยท Bulletproof Infrastructure

TI-2026-013 โ€” "The Krane Botnet" โ€” 32 IPs, 16 Countries, 3 Actor Clusters, One Bulletproof Hosting Ecosystem

A coordinated multi-actor campaign combining a Mirai-family Go SSH scanner (Krane botnet), the mdrfckr persistent backdoor, and a Scaleway-hosted libssh scanner fleet โ€” all operating through a newly assembled bulletproof hosting ecosystem. WHITELABEL-MNT (Daniel Mishayev) links Seychelles shells, Kentucky LLCs, and STARK-INDUSTRIES routing through jurisdictional arbitrage. The campaign targets database administrators (postgres, oracle, mysql), DevOps engineers (git, zabbix, odoo, hadoop), and cloud platforms. Post-exploitation profiles CPU and GPU for crypto-mining viability, implying capability beyond Monero-only deployment. Infrastructure was assembled in January 2026; four newly registered ASNs were simultaneously modified on 2026-05-13, shared Windows RDP VM deployment patterns appear across three ASNs, and C2 staging lives on FranTech bulletproof hosting. One Krane Monero wallet is identified. Across the observed window, the fleet generated 8,854 login attempts and 303 successful authentications. This is not a self-propagating worm โ€” it is a tightly controlled fleet with centralized IP management.
๐Ÿฆ  Krane/mdrfckr Botnet SSH-2.0-Go Scanner 32 IPs ยท 16 Countries ยท 20 ASNs 3 Actor Clusters Identified WHITELABEL-MNT Infrastructure STARK-INDUSTRIES Routing Bulletproof Hosting Ecosystem Crypto-Mining (XMRig/HashVault) Active Campaign ยท 55+ Days ยท Ongoing

๐Ÿ“‹ Executive Summary

Between 2026-03-24 and 2026-05-19 (55+ days, ongoing), a coordinated credential-stuffing operation targeting Linux servers with database and DevOps credentials deployed 32 confirmed IPs across 16 countries and 20 autonomous systems. The campaign uses three distinct toolsets, at minimum two threat actor clusters, and infrastructure tied to a new bulletproof hosting web anchored on WHITELABEL-MNT.

Krane is not simply another SSH scanner. It is a Mirai-family derivative with a Go-based access phase, a multi-architecture tux.* dropper chain, a confirmed XMRig fork miner, and explicit credential targeting for Solana validators. Running in parallel is the mdrfckr backdoor workflow documented in TI-012, reusing the same RSA key, the same three-step authentication pattern, and the same host-scoring logic.

  • 8,854 login attempts produced 303 successful authentications
  • C2 and staging infrastructure spans FranTech/PonyNET, xTom Germany, and Westcall
  • GPU enumeration indicates the operator is scoring hosts for higher-yield mining workloads, not just opportunistic CPU abuse
  • Correlation against 5,300+ honeypot IPs shows a centrally managed fleet, not uncontrolled worm propagation
๐Ÿฏ Honeypot Collection โ€” 2026-03-24 โ†’ 2026-05-19
32
Confirmed IPs
8,854
Login Attempts
20
Autonomous Systems
16
Countries
3
Actor Clusters

The observed fleet mixes scanner nodes, key-injection nodes, relays, and staging infrastructure. Operationally, the interesting number is not just 32 IPs โ€” it is the combination of 303 successful logins, 69% post-auth command execution, and tight infrastructure reuse.

MetricValue
Campaign window2026-03-24 23:21Z โ†’ 2026-05-19 02:37Z
Successful authentications303 of 8,854 attempts (3.4%)
IPs issuing post-auth commands22 of 32 (69%)
Behavioral variants5 profiles: heavy profiler, quick profiler, key injector, fixed dictionary, silent auth
Targeting biasDatabase admins, DevOps/SRE accounts, cloud service users, Solana validators
Observed strategyCredential stuffing โ†’ system profiling โ†’ environment validation โ†’ miner deployment
Campaign structureControlled fleet; no evidence of self-propagating autonomous spread
Key analytical finding

Correlation analysis across the full honeypot corpus found no independent expansion wave carrying the full postgres/oracle/zabbix/odoo/hadoop fingerprint. The operator is choosing and managing infrastructure deliberately โ€” a significant distinction from opportunistic worm-like SSH malware.

The campaign resolves into three operational clusters: a Mirai-Go deployment arm, a persistent backdoor arm, and a separate or semi-separate Scaleway libssh arm. Shared infrastructure and shared post-auth logic indicate overlap, but the tradecraft is not identical across all nodes.

Cluster A ยท 12 IPs

๐Ÿฆ  Mirai-Go โ€œKraneโ€ Scanner

Tool: SSH-2.0-Go
HASSH: 16443846184eafde36765c9bab2f4397
Payloads: tux.x86, tux.mips, tux.arm, tux.ppc, tux.m68k, tux.sh4
Miner: vltrig v6.25.0.4 (XMRig fork) via HashVault
C2: 88.214.20.14, 89.190.156.19, minecraftpixelger39clone.dedyn.io
Special targeting: Solana validator credentials (sol, solv, solana, raydium, anza, testnet)
Wallet: 46qJM6LUpYjVPVS6CYRx3x9HpXboYd3gK5HBd9UgWtCGcrNQJKgTxGJ1TcndT3SLDNFnJvjo8LzjpX2uZL7aEtZiFrsGTLa
Cluster B ยท 6 IPs

๐Ÿ”‘ mdrfckr Persistent Backdoor

Tool: SSH-2.0-libssh_0.12.0
HASSH: af8223ac9914f509afdadfaf5f7ee94e
Key SHA-256: a8460f446be540410004b1a8db4083773fa46f7fe76fa84219c93daa1669f8f2
Root password: y6ekwUbyoOkk
Fingerprint: lockr -ia .ssh
Three-step auth:
345gs5662d34:345gs5662d34 โ†’ fail
root:3245gs5662d34 โ†’ success
payload โ†’ key install / password reset / rival eviction
Cluster C ยท 6 IPs

โ˜๏ธ Scaleway libssh Fleet

Tool: SSH-2.0-libssh_0.11.1
HASSH: 03a80b21afa810682a776a7d42e5e6fb
Infrastructure: 6 observed IPs, all in AS12876
Behavior: fixed 29-attempt dictionary, 10 successes per IP
Interpretation: scripted, highly regular, likely a separate operator or compromised cloud nodes pressed into service
ClusterPrimary ObjectiveInfrastructure SignatureConfidence
Krane / Mirai-GoInitial access + miner stagingGo SSH client, tux.*, vltrig, Solana targetingMEDIUM-HIGH
mdrfckrPersistence + host takeoverRSA key injection, password reset, rival displacementHIGH
Scaleway libsshHigh-volume scripted stuffingAS12876-only, exactly 29 attempts / 10 successesLOW-MEDIUM
Attribution note: the evidence strongly supports shared infrastructure brokerage, but not necessarily a single human keyboard. What is clear is a coordinated ecosystem: one broker layer, multiple access workflows, one monetization model.

Observed behavior resolves into four phases. The miner itself was not captured in the honeypot, but the staging logic, profiling commands, and infrastructure pivot points are sufficient to map the intrusion chain with high confidence.

โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”    โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”    โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”    โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”
โ”‚ PHASE 1     โ”‚โ”€โ”€โ”€โ–ถโ”‚ PHASE 2      โ”‚โ”€โ”€โ”€โ–ถโ”‚ PHASE 3         โ”‚โ”€โ”€โ”€โ–ถโ”‚ PHASE 4      โ”‚
โ”‚ Credential  โ”‚    โ”‚ System       โ”‚    โ”‚ Environment     โ”‚    โ”‚ Miner        โ”‚
โ”‚ Stuffing    โ”‚    โ”‚ Profiling    โ”‚    โ”‚ Validation      โ”‚    โ”‚ Deployment   โ”‚
โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜    โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜    โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜    โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜
  SSH brute-force    uname, nproc,      writable dir test,     XMRig fork
  DB/DevOps creds    CPU model, GPU,    BusyBox/container      (not captured
  Go/libssh tools    lspci nvidia       detection, PATH        in honeypot)
Technique IDNameEvidence
T1110.004Credential Stuffing8,854 attempts using database- and service-account-heavy dictionaries
T1082System Information Discoveryuname, nproc, /proc/cpuinfo, GPU checks
T1083File and Directory DiscoveryWritable directory loops, ls --help for BusyBox/container clues
T1496Resource HijackingXMRig fork (vltrig), HashVault, host scoring for compute value
T1016System Network Configuration Discoveryipinfo.io geolocation checks
T1518Software Discoverylspci | grep -i nvidia, CPU model extraction
T1497.001Virtualization / Sandbox EvasionBusyBox detection, overlay filesystem heuristics, lightweight container checks
T1098.004SSH Authorized Keysmdrfckr RSA public key injection
T1562.004Disable or Modify FirewallFirewall process termination and cleanup behavior
T1059.004Unix ShellAll post-auth activity delivered via shell commands
Five behavioral variants
Heavy profiler โ€” OMEGATECH nodes pull CPU, GPU, writable-dir, and ipinfo.io telemetry
Quick profiler โ€” Linode fleet runs uname + nproc then moves on
Key injector โ€” mdrfckr nodes install persistence, reset passwords, and evict rivals
Fixed dictionary โ€” Scaleway nodes execute exactly 29 attempts and 10 successes
Silent auth โ€” authenticate only, likely validating reachability or stolen creds
Critical implication

GPU checks are the tell. Commodity SSH botnets usually stop at CPU, disk, and cores. Krane explicitly looks for NVIDIA-capable hardware, suggesting target valuation for multi-coin mining or mixed CPU/GPU monetization.

The infrastructure layer is the most consequential part of this investigation. The visible campaign IPs are only the edge. Behind them sits a newly assembled hosting ecosystem built for legal distance, abuse friction, and rapid replacement.

WHITELABEL-MNT broker diagram
                        WHITELABEL-MNT
                       (Daniel Mishayev)
                       DE address, IL code
                              โ”‚
                โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”
                โ–ผ                             โ–ผ
         OMEGATECH                    StormIndustries
         AS202412                     AS214472
         Seychelles shell             "Offshore LC"
         Created Jan 2026             Kentucky LLC #1375656
         Turkey physical
                                              โ”‚
                                              โ–ผ
         GhostyNetworks  โ”€โ”€โ”€routesโ”€โ”€โ–ถ  STARK-INDUSTRIES
         AS205759                      AS44477
         Kentucky LLC #1531972         Ivan Neculiti
         Created Jan 2026              Moldova/Russia, EU-sanctioned
DateEventEntity
2026-01-05Organization createdOmegatech LTD (Seychelles)
2026-01-12ASN assignedAS202412 OMEGATECH
2026-01-16Organization createdGhosty Networks LLC (Kentucky)
2026-01-20ASN assignedAS205759 GHOSTYNETWORKS
2026-05-13Simultaneous RIPE modificationOMEGATECH, GhostyNetworks, StormIndustries, W1N
Jurisdiction arbitrage
EntityChain
OmegatechSeychelles โ†’ Turkey โ†’ Daniel Mishayev (DE/IL)
Ghosty NetworksKentucky โ†’ Netherlands โ†’ STARK-INDUSTRIES (RU)
W1NDoncaster, UK โ†’ unknown hosting โ†’ Ukrainian phone operator
FastnetSeychelles โ†’ USA โ†’ Multacom
C2 and staging nodes
EndpointPurpose
88.214.20.14Mirai dropper host (sshbins.sh)
89.190.156.19Krane botnet C2
staging.bargainejuice.xyzFranTech panel with HTTP 401 Basic Auth
minecraftpixelger39clone.dedyn.ioModule / update endpoint
179.61.185.154:80DTunnel proxy relay (Akronic)
Windows RDP VM deployment pattern
IPASNHostname
91.92.240.199OMEGATECHWIN-4EOAP4JTJR9
45.66.228.255PLAY2GOVM-5CE35281-C77
185.246.128.170W1NWIN-5D3F54F70G8

Three ASNs, one VM deployment style. That is consistent with one operator image or one shared provider template.

ISC SANS abuse telemetry
IPReportsAttacks
156.238.252.13316,368182
185.246.128.1703,90343
91.92.240.199โ€”โ€”
198.98.62.211โ€”โ€”
Why the hosting layer matters

Krane is not just an intrusion set; it is an ecosystem. Seychelles shells, Kentucky LLCs, sanctioned-route transit, and bulletproof hosts allow one campaign to move between jurisdictions faster than most abuse processes can keep up.

The credential dictionary reveals the intended victims more clearly than any geolocation map. This is a technology-led campaign targeting high-value administrative roles, service accounts, and compute-rich Linux systems.

SectorTargeted CredentialsRiskWhy It Matters
Database Administrationpostgres, oracle, mysqlCRITICALDirect access to crown-jewel data stores
DevOps / SREgit, zabbix, odoo, hadoop, jenkinsHIGHInfrastructure pivoting and supply-chain leverage
Cloud / Platformubuntu, admin, rootHIGHFull-system compromise of compute nodes and control planes
IoT / Embeddedpi, default accountsMEDIUMFleet expansion and proxy-node recruitment
Sectors at special risk
Government โ€” Oracle/PostgreSQL systems with legacy SSH exposure
Healthcare โ€” monitoring and admin nodes with weak operational passwords
Education โ€” Hadoop and research clusters with spare compute
MSPs / SMEs โ€” Odoo / ERP environments and DevOps jump hosts
Primary victim logic

The operator is not geographically selective. It is compute selective and role selective: anything with SSH, privileged service accounts, and profitable CPU/GPU capacity is fair game.

This campaign shares the mdrfckr RSA key, the same three-step authentication pattern (345gs5662d34), the same hardware-scoring commands, and overlapping infrastructure (W1N AS42237, OMEGATECH AS202412) with the campaign documented in TI-012 โ€” โ€œThe mdrfckr Botnet.โ€

Shared with TI-012
Same RSA key, same auth dye-marker, same competitor displacement, same host scoring, same ecosystem overlaps
Newly mapped here
The Krane Go scanner, the tux.* Mirai dropper, explicit Solana targeting, and the full WHITELABEL-MNT infrastructure web

TI-012 exposed the enduring backdoor. TI-013 exposes the additional access-and-mining layer wrapped around it โ€” turning what looked like a single botnet into a broader, brokered multi-actor campaign.

The full observed fleet is reproduced below because cluster-level summaries hide the real operational shape: cloud scanners, broker-linked ASNs, staging nodes, and relay infrastructure all coexist inside one campaign picture.

IPCountryASNOrgRole
91.92.240.199TR202412OMEGATECHHeavy profiler
91.92.243.116TR202412OMEGATECHHeavy profiler
172.104.31.246US63949Akamai-LinodeScanner fleet
9.223.113.24GB8075MicrosoftScanner
46.151.182.2NL205759GhostyNetworksScanner fleet
104.105.64.198US63949Akamai-LinodeScanner fleet
66.228.34.48US63949Akamai-LinodeScanner fleet
176.65.139.103DE214472StormIndustriesScanner fleet
50.116.54.193US63949Akamai-LinodeScanner fleet
173.255.229.219US63949Akamai-LinodeScanner fleet
66.175.212.125US63949Akamai-LinodeScanner fleet
209.99.186.189KN402253SKN TelecomScanner
179.61.185.154AE199829Akronic (DTunnel)Proxy relay
34.132.194.42US396982Google CloudScanner
216.48.191.18IN132420E2E NetworksScanner
185.246.128.170GB42237W1NKey injector
156.238.252.133SC8796Fastnet/MultacomScanner
171.25.158.47SE35100PATRIKWEBScanner
198.98.62.211US53667FranTech/PonyNETC2 staging
186.251.71.202BR262826PW InformaticaScanner
103.234.53.173HK40065CNSERVERSScanner
49.229.102.187TH45458SBN-ISPScanner
165.154.5.249HK135377UCloud HKScanner
41.208.147.131SN8346SonatelScanner
163.172.139.67FR12876Scalewaylibssh scanner
151.115.164.231FR12876Scalewaylibssh scanner
151.115.167.143FR12876Scalewaylibssh scanner
151.115.85.114FR12876Scalewaylibssh scanner
51.15.80.78FR12876Scalewaylibssh scanner
51.158.238.156FR12876Scalewaylibssh scanner
51.159.147.158FR12876Scalewaylibssh scanner
45.66.228.255BG215439PLAY2GO-NETScanner
Network IOCs
IOCContext
88.214.20.14Mirai dropper hosting
89.190.156.19Krane botnet C2
198.98.62.211C2 staging / panel host
staging.bargainejuice.xyzC2 panel (numbered subdomains also observed)
minecraftpixelger39clone.dedyn.ioModule / update endpoint
Host-based IOCs
IndicatorValue
HASSH16443846184eafdeโ€ฆ, af8223ac9914f509โ€ฆ, 03a80b21afa81068โ€ฆ, 2ec37a7cc8daf20bโ€ฆ
XMR wallet46qJM6LUpYjVPVS6CYRx3x9HpXboYd3gK5HBd9UgWtCGcrNQJKgTxGJ1TcndT3SLDNFnJvjo8LzjpX2uZL7aEtZiFrsGTLa
Key SHA-256a8460f446be540410004b1a8db4083773fa46f7fe76fa84219c93daa1669f8f2
Dropped filestux.x86, tux.mips, tux.arm, sshbins.sh, vltrig
Credentialsy6ekwUbyoOkk, 345gs5662d34, 3245gs5662d34
Unique commandlockr -ia .ssh

Network-level HASSH detection is useful, but insufficient on its own. The strongest detections combine SSH fingerprinting, credential context, host telemetry, and miner-staging artifacts.

Suricata
alert ssh any any -> any 22 (msg:"KRANE Botnet - Go SSH Scanner HASSH";
  ssh.hassh:"16443846184eafde36765c9bab2f4397"; sid:2026001; rev:1;)

alert ssh any any -> any 22 (msg:"mdrfckr Backdoor - libssh_0.12.0 HASSH";
  ssh.hassh:"af8223ac9914f509afdadfaf5f7ee94e"; sid:2026002; rev:1;)

alert http any any -> any any (msg:"KRANE Botnet - Dropper Download";
  content:"GET"; http_method; content:"/sshbins.sh"; http_uri;
  sid:2026003; rev:1;)

alert dns any any -> any any (msg:"KRANE Botnet - C2 DNS Query";
  dns.query; content:"minecraftpixelger39clone.dedyn.io"; sid:2026004; rev:1;)
YARA
rule mdrfckr_key_injection {
  strings:
    $key = "mdrfckr" ascii
    $lockr = "lockr -ia .ssh" ascii
    $chpasswd = "y6ekwUbyoOkk" ascii
  condition:
    any of them
}
Sigma
title: Crypto-Mining Botnet System Profiling
logsource:
  product: linux
  category: process_creation

detection:
  selection:
    CommandLine|contains:
      - 'lspci | grep -i nvidia'
      - 'lspci | grep -i vga'
      - 'cat /proc/cpuinfo | grep name | wc -l'
      - 'nproc'
  condition: selection
Defender guidance

Best detection stack: HASSH + SSH banner + process telemetry + DNS to dedyn.io / bargainejuice.xyz + hunt for the mdrfckr RSA key in authorized_keys. Any one signal can be noisy; the combination is not.

This dossier blends direct honeypot observation with infrastructure OSINT. The strongest conclusions come from places where session replay, HASSH clustering, RDAP/RIPE ownership data, and public abuse telemetry all align.

Sources
  • Cowrie SSH honeypot (5,300+ observed IPs)
  • RDAP / RIPE NCC registry data
  • AbuseIPDB and ISC SANS DShield telemetry
  • Shodan, PeeringDB, and published honeypot research
  • Session replay, credential clustering, and HASSH correlation
Limitations
  • No miner binary was captured inside the honeypot
  • FranTech C2 staging was not actively authenticated into
  • Diicot overlap remains suggestive, not confirmed
  • Operator nationality cannot be stated with confidence
  • Scaleway cluster may be a separate operator or rented / compromised nodes
FindingConfidence
Campaign is crypto-mining recruitmentHIGH
Three distinct actor clusters are presentHIGH
WHITELABEL-MNT is the infrastructure broker layerHIGH
Krane is a Mirai-family derivativeHIGH
Daniel Mishayev is central to the infrastructure webMEDIUM-HIGH
Scaleway cluster is a separate operatorMEDIUM
Specific human operator nationalityLOW

๐Ÿง  Closing Assessment

Krane is the missing upper layer on top of the mdrfckr story. TI-012 showed the persistence mechanism. TI-013 shows the access pipeline, the miner staging, the Solana-aware credential targeting, and the brokered hosting web that ties the operation together.

The strongest signal in this dossier is not one wallet or one IP. It is the combination of shared credentials, shared post-auth tradecraft, shared broker infrastructure, and synchronized ASN administration. That is what transforms a pile of scanners into a campaign.

Bottom line: defenders should treat this as an active multi-actor crypto-mining ecosystem targeting administrative Linux access at global scale. If your estate still allows SSH password authentication for database or DevOps accounts, this campaign is built to find you.

Investigation Period: 2026-03-24 โ†’ 2026-05-19 (55+ days, ongoing) ยท Classification: TLP:WHITE โ€” Public ยท Data Source: Cowrie SSH honeypot + infrastructure OSINT ยท Core references: RIPE / RDAP, AbuseIPDB, ISC SANS, Shodan, PeeringDB, published honeypot research ยท Assessment focus: credential stuffing, post-auth profiling, infrastructure brokerage, and crypto-mining monetization.
โš  Personal capacity. Research published independently โ€” not reflecting employer views. Derived from passive observation of attacks against personal infrastructure. Full disclaimer โ†’
โ† Previous Honeypot Field Reports โ€” 14 / 20 Next โ†’