The Food Chain
TI-2026-050D ยท Letter 4 of 5 ยท Page 300Step back far enough and the individual campaigns dissolve into something larger. Not fifteen separate scanning operations. One ecosystem. A food chain where each tier feeds the next, information flows upward, and the smallest groups extract the most value.
We've spent three letters inside the Go_SSH fleet โ its tools, its post-authentication behavior, its offshore shells. Now we zoom out. The Go_SSH fleet is one species in an ecosystem of fifteen. Understanding where it sits in the hierarchy reveals not just what it does, but what it is for.
I. The Census of Campaigns
The honeypot identifies fifteen distinct SSH scanning campaigns by HASSH fingerprint, operating simultaneously in the same observation window. Together they command 4,619 IPs, generate 133,428 connection attempts, and achieve 11,500 successful authentications.
| Campaign | Tool | IPs | Attempts | Success | Rate | Actor |
|---|---|---|---|---|---|---|
| libssh 0.11.x | C | 1,313 | 44,341 | 38 | 0.09% | actor-6285990cc704 |
| OpenSSH 7.4 | C | 757 | 2,560 | 59 | 2.30% | โ |
| libssh2 1.11.0 | C | 628 | 16,253 | 0 | 0.00% | actor-0d9fdff8bb3d |
| libssh 0.12.0 | C | 607 | 26,163 | 28 | 0.11% | actor-6285990cc704 |
| libssh 0.9.6 | C | 421 | 14,222 | 24 | 0.17% | actor-6285990cc704 |
| Go_SSH | Go | 320 | 7,115 | 333 | 4.68% | โ |
| AsyncSSH 2.1.0 | Python | 92 | 4,044 | 526 | 13.01% | โ |
| Paramiko | Python | 63 | 67 | 67 | 100% | โ |
| PuTTY | C | 54 | 3,541 | 103 | 2.91% | โ |
| OpenSSH 8.x | C | 45 | 717 | 49 | 6.83% | actor-e00eca62479f |
| Ncrack | C | 34 | 29 | 29 | 100% | โ |
| libssh 0.10.x | C | 34 | 37 | 37 | 100% | โ |
| libssh2 1.11.1 | C | 31 | 4,104 | 107 | 2.61% | โ |
| Hydra | C | 29 | 29 | 29 | 100% | โ |
| openssh_legacy | C | 25 | 2 | 2 | 100% | โ |
Fifteen tools. Four programming languages (C, Go, Python, plus compiled binaries). Four confirmed actor clusters. Twenty-three unattributed campaigns. This is not a single threat. This is an industry.
II. The Three Tiers
Sort the campaigns by success rate and three tiers emerge โ each with a distinct role in the ecosystem's value chain.
These never fail. Every connection is a successful authentication. They target only hosts where the credentials are already known. These are the consumers of the ecosystem โ they exploit what others discover.
Moderate success rates. These test credentials at scale but with some targeting. They sit between the mass scanners and the precision exploiters โ validating which credentials work before exploitation tools arrive.
These are the foundation. 4,046 IPs spray credentials and enumerate SSH services across the entire internet. Most attempts fail. That's the point โ mass scanning is information gathering, not exploitation. The 0.44% aggregate success rate is the ore. The tiers above refine it.
The Value Chain
Tier 1 identifies which hosts run SSH. Tier 2 determines which hosts accept default credentials. Tier 3 deploys payloads on confirmed targets. Whether these tiers collaborate directly (shared databases, API access, commercial relationships) or independently (dark web credential markets, access broker platforms), the information flows in one direction: upward. Each tier adds value. Each tier extracts a cut. The result is an industrial process that turns internet-wide scanning into monetized access.
Where Go_SSH Sits
Go_SSH at 4.68% success occupies Tier 2 โ the validation layer. This position explains everything we've observed across the previous letters:
โข The two-phase architecture (050B) makes sense as a Tier 2 operation: validate credentials at scale, then profile validated hosts for Tier 3 consumption.
โข The offshore shell infrastructure (050C) makes sense as persistent Tier 2 infrastructure: the TechTies recon nodes are the bridge between validation (Tier 2) and exploitation (Tier 3).
โข The mega-command's GPU detection (050B) makes sense as Tier 2 triage: GPU-equipped hosts are flagged as high-value for Tier 3 exploiters who deploy mining software.
โข The honeypot detection (050B) makes sense as Tier 2 quality control: remove false positives before passing the inventory upward.
Go_SSH is not the endgame. It is the supply chain.
III. The Four Actors
Of fifteen campaigns, four have been linked to confirmed actor clusters by the automated attribution system. Each cluster represents a distinct operational identity.
The libssh Swarm
3 campaigns ยท 2,341 IPs ยท 84 countries
84,726 attempts ยท 90 successes
Tier 1: Mass Scanner
Asian cloud infrastructure
The libssh2 Recon Fleet
1 campaign ยท 628 IPs ยท 48 countries
16,253 attempts ยท 0 successes
Tier 1: Pure Reconnaissance
Service enumeration only
The OpenSSH 8.x Fleet
1 campaign ยท 45 IPs ยท 15 countries
717 attempts ยท 49 successes
Tier 2: Targeted Validation
6.83% success rate
Unknown
Details minimal
Appears in attribution data
Role uncertain
Likely smaller operation
Four confirmed. Eleven unattributed โ including Go_SSH, AsyncSSH, Paramiko, PuTTY, Ncrack, and Hydra. The attribution system links campaigns when they share IPs, credentials, command fingerprints, or infrastructure patterns. The Go_SSH fleet shares none of these with any attributed cluster โ confirming our finding from 050A that it operates independently.
But "independent" does not mean "unconnected." The TechTies bridge links Go_SSH infrastructure to the libssh swarm's actor-6285990cc704. The ecosystem is connected by infrastructure, not by operations.
IV. The Clock
The Go_SSH fleet's temporal pattern is unlike any other campaign in the ecosystem.
03:00 โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ 1,108
04:00 โโโโโโโโโโโโโโ 292
06:00 โโโโโ 119
07:00 โโโโ 96
09:00 โโโโ 89
11:00 โโโโ 88
14:00 โโโโ 87
All other hours combined: ~572
Peak window: 02:00โ04:00 UTC = 2,133 connections (61.4%)
Sixty-one percent of all Go_SSH activity occurs in a two-hour window: 02:00 to 04:00 UTC. This is not random scanning. This is a scheduled batch job.
The Timezone Translation
02:00-04:00 UTC translates to:
05:00โ07:00 EET (Bucharest, Helsinki, Athens) โ early morning
03:00โ05:00 GMT (London) โ deep night
07:30โ09:30 IST (Mumbai) โ morning rush
10:00โ12:00 CST (Beijing) โ mid-morning
The CET/EET window is the most compelling: 04:00-06:00 CET is when European hosting providers' abuse teams are asleep, automated monitoring systems are processing overnight logs, and network traffic is at its daily minimum. For a European operator, this is the optimal window to run scanning batches โ maximum activity, minimum observation.
Compare with the libssh swarm from TI-2026-049E: the libssh burst was also 02:00-04:00 UTC, but interpreted as 10:00-12:00 CST (China Standard Time) โ mid-morning for Chinese operators. Same UTC window, different local interpretation. The timestamp alone cannot distinguish European pre-dawn from Chinese mid-morning.
The Wednesday Anomaly
78.5% of weekly Go_SSH activity falls on Wednesday (2,729 of 3,476 connections). This extreme concentration rules out continuous automated scanning and confirms human scheduling. The operator runs batch scanning jobs, and Wednesday is the primary execution day. Why Wednesday? Possibly: mid-week means 4 days before weekend abuse-team response windows. Or: the operator has a weekly operational cycle (Monday: acquire IPs, Tuesday: prepare target lists, Wednesday: execute scans, Thursday-Friday: triage results, Weekend: off).
V. The Third Shell
While mapping the broader ecosystem, a third Seychelles entity emerged: IP Volume inc (AS202425).
Omegatech LTD ยท AS202412 ยท RDAP: NL
IP Volume inc ยท AS202425 ยท INT-NETWORK ยท RDAP: NL
All three: Seychelles registration ยท Netherlands RDAP ยท abuse score 100
Shared IP prefix: 91.92.x.x (TechTies + Omegatech confirmed)
Three separate ASNs โ three independent takedown targets
Three shell companies. Three ASNs. One jurisdiction. If any single entity is suspended by RIPE (the European IP registry), the other two continue operating with their separate ASN allocations and separate RDAP records. This is infrastructure resilience through corporate multiplication โ the same principle that drives financial shell company networks, applied to network infrastructure.
The Question Nobody Is Asking
RIPE NCC allocates ASNs to entities after membership verification. Three Seychelles-registered companies โ all with 100% abuse scores, all with NL RDAP, all sharing IP prefix space โ have been assigned three separate ASNs by the same Regional Internet Registry. Either RIPE's due diligence process cannot detect shell company clusters, or the entities met the formal requirements and the abuse pattern developed after allocation. Either way, three ASNs are now operational infrastructure for scanning campaigns and there is no mechanism to revoke them as a group.
VI. The Four Downloads
Buried in the timeline data: four file downloads. In 3,476 connections over seven days, four resulted in a file being downloaded to the honeypot. We had not seen this in earlier session samples because the downloads occurred during the peak activity windows โ lost in the flood of validation sessions.
Four downloads from 333 successful authentications. That's 1.2%. If these represent exploitation payloads โ cryptominers, backdoors, botnet clients โ then only 1.2% of validated hosts receive a payload. The rest remain in inventory: credentials confirmed, host profiled, but not yet exploited. This matches the access broker model: validate thousands, exploit dozens, hold the rest for sale.
The Conversion Funnel
7,115 credential attempts โ 333 successful authentications (4.68%) โ ~12 hosts profiled with mega-command (3.6% of successes) โ 4 payloads deployed (1.2% of successes, 0.06% of all attempts).
This is a conversion funnel with a 0.06% end-to-end rate. For every 10,000 SSH connections the Go_SSH fleet makes, six result in a deployed payload. The funnel is efficient because each stage filters aggressively: bad credentials are discarded, honeypots are detected, low-value hosts are inventoried but not exploited. Only the highest-value targets receive the payload. The operation is selective by design.
VII. The Ecosystem Is the Threat
The individual campaigns โ libssh, Go_SSH, AsyncSSH, Paramiko โ are components. The ecosystem is the threat.
4,619 IPs from 15 campaigns scan the same target space simultaneously. They use different tools, different infrastructure, different strategies. Some are linked by actor clusters. Some share infrastructure through Seychelles shells. Some operate entirely independently. But together, they create a comprehensive, continuous, multi-layered assault on every SSH server connected to the internet.
No single campaign covers every target. But the ecosystem does. No single tool evades every defense. But the diversity of tools ensures that some percentage always gets through. No single operator has the best credentials, the best infrastructure, the best timing. But the ecosystem, collectively, has all of these.
The lesson is not "block Go_SSH." The lesson is that SSH scanning is an economic ecosystem with division of labor, specialization, supply chains, and value extraction at every tier. Disrupting one campaign shifts activity to others. The only effective defense is at the target level: strong credentials, key-based authentication, and monitoring that detects post-compromise behavior regardless of which scanner found the door.
VIII. The Thread Forward
The final letter of this series brings it all together. The Go_SSH fleet. The libssh swarm. The six ecosystems. The three Seychelles shells. The two-phase architecture. The mega-command. The food chain.
Letter E is the synthesis โ the complete picture of what the Toolmakers built, how it operates, and what it means for every SSH server on the internet. It is the assessment. The conclusions. And the recommendations for anyone defending against an ecosystem that never sleeps, even if it only runs its batch jobs on Wednesday mornings.
TI-2026-050D ยท The Food Chain ยท Series: The Toolmakers (4/5)
Classification: TLP:GREEN ยท Confidence: HIGH
Sources: Cowrie honeypot, campaign attribution engine, HASSH fingerprint database, RIPE RDAP, temporal analysis
Cross-references: TI-2026-050C, TI-2026-049E, TI-2026-050B
Published: June 30, 2026 ยท shuffle-on.com