Incident casework · Case 3 of 4

From phishing email to abuse reports: triaging a ScreenConnect MSI campaign

A user forwarded a suspicious email. Before we could look at it, a second user mentioned they had already tried to download whatever it linked to and "it wasn't compatible with Windows." That phrase got my attention fast.

Written March 2026, from an enterprise incident I worked. Published here September 2026.
Anonymized Our organization, the peer organization whose mailbox was compromised, and its location are generalized. Attacker infrastructure and the relay session ID are shown as observed.
Outcome No installation on either exposed endpoint. Five abuse reports filed. One provider unreachable.

Was the sender spoofed or compromised?

Always the first question. The email, subject "Requesting Reviewing Document Via Adobe Cloud," came from a real person at a peer organization we had worked with before. If the account was spoofed, it is an external problem. If it was compromised, someone is sitting in that mailbox right now sending phishing to the entire contact list.

The headers told the whole story:

  • Message-ID ended in @mail.yahoo.com. The email originated from real Yahoo infrastructure, not a random SMTP relay.
  • X-Mailer: WebService/1.1.25380 YMailNorrin. Yahoo Mail web client. Someone logged into this account through a browser and hit send.
  • DKIM signature signed by yahoo.com and valid. This is the key detail. A spoofed message would fail DKIM or be signed by a different domain. A valid Yahoo signature means Yahoo authenticated the sender.
  • MX records for the sender's domain pointed to mx-biz.mail.am0.yahoodns.net, Yahoo Small Business hosting. The domain's mail is Yahoo and the signature is Yahoo. Everything is consistent.

Verdict: compromised, not spoofed. The attacker has this person's credentials and sent from their actual Yahoo-hosted mailbox. DKIM passed because the email genuinely came from where it claims.

The email gateway scored it at -0.82 against a kill threshold of 3.0. DKIM passed. SPF passed. The only flag was a missing To: header, which added +1.21, nowhere near enough. The gateway was never going to catch this because, technically, it is a legitimate email from a legitimate mailbox. It just happens to be sent by an attacker.

The payload

The link went through ctrk.klclick3[.]com, a Klaviyo email-marketing tracking link. Klaviyo is a legitimate platform; the attacker was abusing it for redirect infrastructure. It redirected to dccyop[.]org/[redacted], which served ScreenConnect.ClientSetup.msi, a 16.7 MB installer.

The user who said it "wasn't compatible with Windows" almost certainly hit a SmartScreen warning or a browser download block. That warning did its job.

Sandbox analysis

Sandbox score 8/10. A legitimate ScreenConnect installer, weaponized for unauthorized access:

  • C2 relay: instance-x7bxki-relay.screenconnect[.]com resolving to 15.235.110.89 (OVH Canada)
  • Session ID c74a4df9-a1ae-422f-8ab0-a017da1ee909 hardcoded in the installer config
  • Credential-provider DLL and authentication-package DLL installed, hooking the Windows credential pipeline
  • 64 privilege-escalation attempts: SeDebugPrivilege, SeTcbPrivilege and more
  • 37+ COM-hijacking registry modifications

This is the RMM-abuse pattern that is everywhere right now. The attacker does not write custom malware. They configure a ScreenConnect instance, bake the connection details into the MSI, and get persistent remote access through a signed, legitimate management tool. It installs as a service, survives reboots, and looks like normal IT software to anyone not specifically hunting unauthorized RMM.

Endpoint verification

Two machines received the email. I pulled endpoint trajectory data for both via the API and checked for ScreenConnect artifacts: files, services, registry keys, connections to the relay. Both clean. One user downloaded the MSI but never ran it. The other never clicked.

No installation on either machine. SmartScreen and user caution did the job.

The abuse reporting chain

This is where a lot of IR stops short. We confirmed no compromise on our end, but the attacker's infrastructure is still live. Every hour that relay stays up, more organizations get the same MSI. So I spent about ten minutes filing reports:

  • ConnectWise: the weaponized relay instance with its session ID. ConnectWise has an abuse process and can disable the instance remotely.
  • OVH: the C2 server IP, with full context.
  • APWG: full IOC report to [email protected].
  • Klaviyo: the abused tracking link, to [email protected].
  • The compromised user's organization: contacted directly so they know the account is actively sending phishing.

And then there was Yahoo. Yahoo's abuse reporting is broken. The mailbox is compromised and actively phishing. The online abuse form is non-functional. [email protected] bounced with 554 mailbox not found. One of the largest mail providers in the world and you cannot report a compromised account to them. The attacker keeps sending until the owner notices or Yahoo's automation catches on. I documented the failed attempts and moved on.

Key takeaways

DKIM tells you compromised vs. spoofed. Valid signature, signing domain matching the sender's mail infrastructure: someone has the credentials. Failed or foreign-domain signature: spoofed. The distinction changes your response, because compromised means the sender's organization needs to hear from you immediately.

Gateways cannot catch compromised-account phishing. Scored at -0.82, DKIM and SPF passing. User awareness is the only layer left.

RMM abuse is the dominant delivery method. ScreenConnect, AnyDesk, Action1: signed, trusted, blends in. If you are not explicitly blocking unauthorized RMM tools, you are relying on users not to run the installer.

Report upstream. ConnectWise can disable the relay. OVH can null-route the IP. Klaviyo can kill the redirect. Ten minutes, and it disrupts the tooling for everyone downstream of you.

IOCs

TypeValue
Redirectctrk.klclick3[.]com (Klaviyo, abused)
Payload hostdccyop[.]org
C2 relayinstance-x7bxki-relay.screenconnect[.]com
C2 IP15.235.110.89 (OVH Canada)
SHA-2561BE180E92DC8F61434CA170203F16DC1CDE5B3AD26FB80A2096C48F04070F9B2
Sender[compromised mailbox at a peer organization]

Tools Endpoint trajectory API · sandbox analysis · email-gateway header analysis and scoring · PowerShell COM automation of Outlook for header extraction

← All research · Next case: a ScreenConnect bat-file attack, and proving nothing left →