ClickFix, CrashFix, InstallFix, FileFix & GhostClaw: Growing Family of Copy-and-Paste Social Engineering Attacks
ClickFix, CrashFix, InstallFix, FileFix & GhostClaw (TL-2026-1551), also tracked as ClickFix, is a high-severity social-engineering campaign, first published 2026-07-19. It has no confirmed attribution, affects Microsoft Windows (PowerShell, Run dialog, File Explorer), maps to 42 MITRE ATT&CK techniques (T1005, T1016, T1027), and is covered by 9 detection rules and 30 indicators of compromise.
Key facts for TL-2026-1551
- Threat ID
- TL-2026-1551
- Also known as
- ClickFix, CrashFix, InstallFix, FileFix, TerminalFix, GhostClaw
- Severity
- HIGH
- Status
- ACTIVE
- Category
- SOCIAL_ENGINEERING
- First published
- 2026-07-19
- Last reviewed
- 2026-07-19
- Attribution confidence
- LOW
- Motivation
- FINANCIAL
- Target sectors
- technology, softwaredevelopment, finance, cryptocurrency, mspmanagedserviceproviders, generalenterprise
- Target regions
- Global
- Detection rules
- 9
- Indicators of compromise
- 30
Malware and tooling in ClickFix, CrashFix, InstallFix, FileFix & GhostClaw
Malware and tooling: AMOS, DigitStealer, GhostClaw, MacSync, ModeloRAT, finger.exe (LOLBin abuse)
A rapidly diversifying family of copy-and-paste social engineering techniques (ClickFix, CrashFix, InstallFix, FileFix, TerminalFix) tricks users into pasting attacker-supplied commands into PowerShell, Run dialogs, macOS Terminal, or the Windows File Explorer address bar via fake CAPTCHA checks, browser-crash prompts, or installer failures — leading to RAT deployment, credential/session theft, and persistence. A related npm supply-chain variant, GhostClaw, impersonated the OpenClaw AI-agent CLI to distribute a cross-platform infostealer/RAT via a malicious dependency.
How ClickFix, CrashFix, InstallFix, FileFix & GhostClaw works
ClickFix-family attacks exploit a consistent psychological pattern: present the victim with a fabricated problem (a broken CAPTCHA, a crashed browser, a failed installer, an inaccessible file) and a deceptively simple 'fix' that requires the victim to paste and run attacker-controlled code themselves, sidestepping browser download protections, mark-of-the-web warnings, and email attachment scanning because the user — not the attacker — executes the payload.
The original ClickFix lure presents a fake 'I'm not a robot' verification page whose 'Fix' button copies a PowerShell or Run-dialog command to the clipboard and instructs the victim to press Win+R and paste it. Huntress reported a 631% increase in ClickFix-related incidents over a six-month period (Aug 2024-Aug 2025), and ClickFix fueled 53% of all malware loader activity observed by Huntress in 2025.
CrashFix (documented by Microsoft, Feb 2026) escalates the lure: a malicious browser extension masquerading as 'uBlock Origin Lite' (distributed via malvertising redirecting to a spoofed Chrome Web Store) deliberately triggers an infinite-loop denial-of-service against the victim's own browser after a delay, then a fake 'CrashFix' repair pop-up pressures the user into pasting a recovery command. The command copies finger.exe to %TEMP% renamed as ct.exe (SHA-256 beb0229043741a7c7bfbb4f39d00f583e37ea378d11ed3302d0a2bc30f267006) to abuse the little-monitored Windows finger utility (TCP/79) as a covert retrieval/beaconing channel, then downloads an obfuscated script.ps1 (SHA-256 c76c0146407069fd4c271d6e1e03448c481f0970ddbe7042b31f552e37b55817) that fingerprints the host for analysis tooling (Wireshark, Process Hacker, WinDbg) and domain-join status before exfiltrating recon data via POST. Domain-joined targets receive a portable WinPython distribution (WPy64-31401) bundling a Python RAT payload (modes.py), identified as ModeloRAT, executed covertly via pythonw.exe. ModeloRAT beacons over HTTP to {C2_IP}:80/beacon/{client_id} and performs domain/network reconnaissance with nltest, whoami, and net use. Persistence is established via an HKCU Run key and a scheduled task named 'SoftwareProtection' that re-executes a Python payload (udp.pyw) every 5 minutes to blend with legitimate Windows Update Medic / software-protection services.
InstallFix and TerminalFix present the copy-paste lure inside a fake failed-installer screen or a macOS/Linux Terminal 'verification' prompt respectively — functionally identical social-engineering wrappers around the same clipboard-hijack mechanic, now cross-platform. Microsoft's parallel Feb-May 2026 reporting ('Infostealers without borders', 'ClickFix campaign uses fake macOS utilities lures') documents ClickFix-style DMG installer and Terminal-paste lures delivering macOS-native and Python-based infostealers — DigitStealer, MacSync, and Atomic macOS Stealer (AMOS) — often bundled together in the same campaign infrastructure, hosted on legitimate platforms (Cloudflare Pages, Squarespace, Tencent EdgeOne) and advertised via malicious search-engine ads impersonating developer tools such as Anthropic's Claude Code. MacSync supports dynamic AppleScript payloads and in-memory execution to evade static/behavioral detection; DigitStealer performs locale/country checks and exits on certain configurations, consistent with organized-crime operational discipline.
FileFix (researcher mr.d0x, disclosed 2025, actively tested in the wild per Check Point) is the most significant structural variant: instead of the Win+R Run dialog, the lure triggers a browser file-upload dialog (which opens Windows File Explorer) and instructs the victim to paste a 'file path' into the Explorer address bar via Ctrl+L. JavaScript on the page silently places a PowerShell command on the clipboard, concatenated after a comment token with a fake, benign-looking file path so the visible pasted text looks legitimate while the full clipboard content executes as a command when the Explorer address bar interprets it. Because the technique lives in File Explorer rather than the Run dialog or a terminal, it evades a large share of user security-awareness training built specifically around 'never paste into Run/PowerShell.' Check Point observed a known ClickFix threat actor (previously targeting cryptocurrency-platform users) testing FileFix with benign payloads within two weeks of public disclosure, signaling operational adoption ahead of live malware delivery.
GhostClaw (JFrog Security Research, discovered/disclosed March 2026) is a supply-chain pivot of the same family into the npm ecosystem: a malicious package, @openclaw-ai/openclawai (versions 1.5.14/1.5.15), impersonated the legitimate 'OpenClaw' AI-agent CLI installer. Uploaded to npm on 2026-03-03 by user 'openclaw-ai', downloaded 178 times before removal around 2026-03-10. A postinstall hook silently re-installs the package globally to place an 'openclaw' binary on PATH. The first-stage dropper, scripts/setup.js, is obfuscated via string-table shuffling, RC4 decoding, and control-flow flattening; it fetches an ~11,700-line second-stage JavaScript bundle (internally named GhostLoader, publicly tracked as GhostClaw) from C2 domain trackpipe[.]dev, decrypted with AES-256-GCM (16-byte IV/16-byte auth tag). The C2 domain and paths are hidden as XOR-encoded integer arrays in source; the bootstrap endpoint is /t/bootstrap?t=<GUID> tagged with campaign ID 'complexarchaeologist1', and the malware performs IP geolocation via ipinfo.io/json. GhostClaw social-engineers the victim's own OS account password (fake Keychain prompt on macOS, PowerShell auth prompt on Windows, su on Linux) to decrypt browser-stored credentials across Chrome, Brave, Edge, Vivaldi, Opera, Yandex, and Comet (via CDP and SQLite extraction), plus Firefox via NSS/PKCS12. On macOS with Full Disk Access it harvests the Keychain databases (login.keychain-db and iCloud Keychain, including -wal/-shm journals), Apple Notes (NoteStore.sqlite), iMessage history (chat.db, with real-time monitoring), Safari history, Mail account configs, and Apple Account data. It targets SSH keys, AWS/Azure/GCP/Kubernetes/Docker credentials, npm/Git/GitHub CLI configs, and a broad set of desktop and browser-extension crypto wallets (Exodus, Electrum, Atomic, Bitcoin Core, Ledger Live, Sparrow, Wasabi, Trezor Suite, MetaMask, Phantom, Solflare), including BIP-39 seed-phrase scanning via wordlist ratio matching, and even scrapes credentials/state for other AI-agent CLIs (ZeroClaw, PicoClaw, OpenClaw). Persistence: on macOS/Linux, monitor.js is dropped to ~/.cache/.npm_telemetry/ and hooked into ~/.zshrc, ~/.bashrc, ~/.bash_profile disguised as '# NPM Telemetry Integration Service' with a PID-based watchdog; Linux adds an @reboot cron entry disguised as '# Node.js Telemetry Collection'; Windows installs to %APPDATA%/.npm_telemetry/monitor.js. Stolen data is archived as tar.gz and exfiltrated primarily to trackpipe.dev, with a Telegram Bot API fallback (<49MB) or GoFile.io with a password-locked, bearer-token-authenticated fallback for larger archives. The RAT component beacons every ~25s (30% jitter) and supports EXEC (shell command, 85s timeout), OPEN (URL), UPDATE (AES-256-GCM-encrypted remote payload), GRAB (file exfil), RECOLLECT (re-run collection), PROXY_START/STOP/STATUS (SOCKS5 proxy), CLONE_START/STOP/STATUS (headless browser cloning via CDP relay), and NUKE (self-destruct); it also monitors the clipboard every 3 seconds for private-key and crypto-address patterns (BTC/ETH, hex/WIF/Solana), AWS keys, and OpenAI/Stripe API keys. Anti-forensics measures include detached child processes (stdio ignored), 60-second temp-payload deletion, PID-lock single-instance enforcement, loot-directory cleanup post-exfiltration, and debug-log exfiltration to Telegram before exit.
Collectively this threat family represents a single evolving tradecraft lineage — 'make the victim run the payload themselves' — expanding across lure surface (CAPTCHA → browser crash → installer → file dialog → terminal), operating system (Windows → macOS → Linux), and distribution channel (malvertising → search ads → npm supply chain), with rising real-world incident volume and confirmed live weaponization of the newest (FileFix, GhostClaw) variants.
MITRE ATT&CK techniques used in TL-2026-1551
Collection
T1005 Data from Local System; T1115 Clipboard Data; T1560.001 Archive via Utility
Discovery
T1016 System Network Configuration Discovery; T1057 Process Discovery; T1082 System Information Discovery; T1482 Domain Trust Discovery; T1518.001 Security Software Discovery
Defense Evasion
T1027 Obfuscated Files or Information; T1036.003 Rename Legitimate Utilities; T1036.005 Match Legitimate Resource Name or Location; T1070.004 File Deletion; T1140 Deobfuscate/Decode Files or Information; T1497.001 System Checks
Exfiltration
T1041 Exfiltration Over C2 Channel; T1567.002 Exfiltration to Cloud Storage
Execution
T1053.003 Cron; T1059.001 PowerShell; T1059.004 Unix Shell; T1059.006 Python; T1059.007 JavaScript; T1204.001 Malicious Link
Persistence
T1053.005 Scheduled Task; T1505 Server Software Component; T1546.004 Unix Shell Configuration Modification; T1547.001 Registry Run Keys / Startup Folder
Privilege Escalation
Credential Access
T1056.003 Web Portal Capture; T1539 Steal Web Session Cookie; T1552.001 Credentials In Files; T1555.001 Keychain; T1555.003 Credentials from Web Browsers
Command and Control
T1071.001 Web Protocols; T1090.002 External Proxy; T1102 Web Service; T1105 Ingress Tool Transfer; T1573.001 Symmetric Cryptography
Initial Access
T1189 Drive-by Compromise; T1195.001 Compromise Software Dependencies and Development Tools
initial-access
T1195.001 Compromise Software Dependencies and Development Tools
Impact
T1499 Endpoint Denial of Service
Resource Development
T1583.001 Domains; T1585 Establish Accounts
defense-impairment
Affected products and versions in ClickFix, CrashFix, InstallFix, FileFix & GhostClaw
- Microsoft — Windows (PowerShell, Run dialog, File Explorer)
Vulnerable versions: Windows 10; Windows 11 - Apple — macOS (Terminal, Keychain, Full Disk Access)
Vulnerable versions: macOS (all supported versions) - npm, Inc. — npm registry package @openclaw-ai/openclawai
Vulnerable versions: 1.5.14; 1.5.15
Fixed in: package removed from registry ~2026-03-10 - Google — Chrome Web Store (fake uBlock Origin Lite extension)
Vulnerable versions: N/A - malicious third-party extension
Remediation for ClickFix, CrashFix, InstallFix, FileFix & GhostClaw
Immediate actions
- Block outbound traffic to trackpipe[.]dev and nexsnield[.]com
- Block TCP/79 (finger protocol) egress at perimeter firewalls
- Remove/quarantine any installs of npm package @openclaw-ai/openclawai (versions 1.5.14/1.5.15)
- Search endpoints for renamed finger.exe artifacts (ct.exe with SHA-256 beb0229043741a7c7bfbb4f39d00f583e37ea378d11ed3302d0a2bc30f267006)
- Remove scheduled task 'SoftwareProtection' and inspect HKCU\Software\Microsoft\Windows\CurrentVersion\Run for unauthorized entries
- Audit ~/.zshrc, ~/.bashrc, ~/.bash_profile and crontab -l for entries referencing 'NPM Telemetry' or 'Node.js Telemetry Collection'
Workarounds
- Disable browser password syncing/autofill decryption on managed devices via Group Policy
- Restrict clipboard access permissions for untrusted browser extensions
Longer-term hardening
- Deploy EDR/behavioral detection for PowerShell processes spawned by browser processes (explorer.exe, chrome.exe, msedge.exe as parent)
- Disable or restrict the Windows Run dialog (Win+R) via GPO where feasible for high-risk user populations
- Deploy npm supply-chain scanning (typosquat/impersonation detection) in CI/CD and developer endpoints
- Expand security-awareness training beyond Run-dialog paste warnings to cover File Explorer address-bar paste and Terminal paste vectors
- Enable Full Disk Access auditing on managed macOS fleets to detect unauthorized grants
Weaknesses (CWE) in ClickFix, CrashFix, InstallFix, FileFix & GhostClaw
CWE-451, CWE-506, CWE-1021
Timeline of ClickFix, CrashFix, InstallFix, FileFix & GhostClaw
- Researcher mr.d0x publicly discloses the FileFix technique, a ClickFix alternative abusing the Windows File Explorer address bar and browser file-upload dialog to execute hidden PowerShell commands.
- Start of the six-month window (Aug 2024-Aug 2025) over which Huntress measured a 631% increase in ClickFix-related incidents, with ClickFix accounting for 53% of all malware loader activity observed in 2025.
- Check Point Research observes a known ClickFix threat actor, previously targeting cryptocurrency-platform users, actively testing the FileFix technique in the wild with benign payloads within two weeks of disclosure gaining traction.
- Microsoft publishes 'Infostealers without borders,' documenting macOS-targeted ClickFix-style social engineering delivering Python-based and native macOS infostealers since late 2025.
- Microsoft Defender Experts disclose CrashFix: a ClickFix variant using a fake malicious 'uBlock Origin Lite' browser extension to deliberately crash the victim's browser via an infinite loop, then socially engineers the user into deploying a portable-Python ModeloRAT payload via renamed finger.exe (ct.exe) and obfuscated PowerShell.
- Reporting (TheHackerNews, Microsoft) documents ClickFix campaigns spreading the MacSync macOS infostealer via fake AI-tool installers, frequently co-deployed with DigitStealer and Atomic macOS Stealer (AMOS) on shared infrastructure.
- JFrog Security Research publishes 'GhostClaw Unmasked,' a full technical analysis of the malicious package's infection chain, C2 infrastructure (trackpipe[.]dev), and cross-platform data-theft/RAT capabilities.
- The malicious npm package @openclaw-ai/openclawai (impersonating the OpenClaw AI-agent CLI) is uploaded to the npm registry under user 'openclaw-ai', beginning distribution of the GhostClaw/GhostLoader cross-platform infostealer and RAT.
- The @openclaw-ai/openclawai package is removed from the npm registry after 178 downloads, following identification by JFrog Security Research.
- Microsoft publishes further reporting on ClickFix campaigns using fake macOS utility lures hosted on legitimate platforms (Cloudflare Pages, Squarespace, Tencent EdgeOne) and distributed via malicious search ads impersonating developer tools including Anthropic's Claude Code.
- Pen Test Partners publishes a consolidated overview of the ClickFix/CrashFix/InstallFix/FileFix/GhostClaw family, framing it as a single growing tradecraft lineage across lure surface, OS, and distribution channel — the primary source for this threat record.
Sources cited for ClickFix, CrashFix, InstallFix, FileFix & GhostClaw
- ClickFix, CrashFix and the Growing Family of Copy and Paste Attacks
- New Clickfix variant 'CrashFix' deploying Python Remote Access Trojan
- Infostealers without borders: macOS, Python stealers, and platform abuse
- ClickFix campaign uses fake macOS utilities lures to deliver infostealers
- GhostClaw Unmasked: A Malicious npm Package Impersonating OpenClaw to Steal Everything
- Malicious npm Package Posing as OpenClaw Installer Deploys RAT, Steals macOS Credentials
- ClickFix Attack: Variants, Detection & How It Works
- FileFix - A ClickFix Alternative
- FileFix: The New Social Engineering Attack Building on ClickFix Tested in the Wild
- ClickFix Campaigns Spread MacSync macOS Infostealer via Fake AI Tool Installers
- Microsoft Discloses DNS-Based ClickFix Attack Using Nslookup for Malware Staging
- New FileFix Attack Weaponizes Windows File Explorer for Stealthy Commands
Threats related to ClickFix, CrashFix, InstallFix, FileFix & GhostClaw
- ClickFix Social-Engineering Technique Becomes Dominant Malware Delivery and Defense-Evasion Vector (CrashFix, FileFix, ConsentFix Variants)
- Mistic Self-Destructing In-Memory Backdoor (MLTBackdoor) Deployed by KongTuke Access Broker Enabling Multi-Group Ransomware Intrusions
- ClickLock Stealer: macOS ClickFix Infostealer Uses 210ms Process-Kill Loops and Fake Authentication Dialogs to Coerce Credentials
- KongTuke ClickFix Campaign — ModeloRAT Deployment via Compromised WordPress Sites and CrashFix Browser Extension
- ClickFix Evolution — nslookup DNS Smuggling + CrashFix Browser DoS + ModeloRAT Python RAT, KongTuke Actor, Enterprise Domain-Joined Targeting
- LabubaRAT: Rust-Based RAT Masquerades as NVIDIA Container Runtime to Backdoor Windows Hosts
Detection coverage for TL-2026-1551
As of 2026-07-19, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-1551 across Splunk SPL, Microsoft KQL and Sigma, covering 30 indicator(s) of compromise. The whole corpus is readable without an account; a free account unlocks full detection query text in Splunk SPL, Microsoft KQL and Sigma; paid tiers add raw indicator values, correlation and the MCP server. Threadlinqs MCP server · View plans.