Skip to main content
Live Webinar with SANS (June 25)— Agentic CTI Automation for Fun & ProfitRegister Free
Mallory
Back to malware
Malware

GhostLoader

GhostLoader is a cross-platform information stealer, also referred to in reporting as GhostClaw, that targets developer environments and trusted development workflows. It has been observed in March 2026 campaigns abusing the OpenClaw AI agent framework, malicious GitHub repositories, shell installers, and a malicious npm package impersonating an OpenClaw installer (@openclaw-ai/openclawai). In the OpenClaw/DeepSeek-Claw campaign, GhostLoader was delivered via obfuscated Node.js payloads embedded in npm lifecycle scripts and manual installation paths such as install.sh or npm install; on macOS and Linux it used terminal-based social engineering and fake password prompts to harvest credentials, while on Windows it could also be delivered through the manual path. Reporting also states the malware internally identifies itself as GhostLoader, while the broader campaign is tracked as GhostClaw.

Documented capabilities include theft of system credentials, browser credentials, cookies, credit cards and autofill data, macOS Keychain and iCloud Keychain databases, SSH keys, cryptocurrency wallet files and seed phrases, cloud and developer credentials and configuration files, API tokens, and other sensitive developer secrets. Multiple sources specifically mention theft of macOS Keychain data, SSH keys, cryptocurrency wallets, cloud API tokens, AWS/Azure/GCP/Kubernetes/Docker/GitHub-related credentials, and browser data. On macOS, if additional permissions are obtained, reporting states it can also access data such as Apple Notes, iMessage history, Safari history, and Mail-related data. Some reporting describes RAT-like functionality including persistence, arbitrary command execution, SOCKS5 proxying, and live browser session cloning via Chrome DevTools Protocol.

Observed behavior includes use of heavily obfuscated JavaScript stages, fake CLI installer interfaces with progress output, fraudulent Keychain-style or sudo-style prompts, validation of entered passwords against legitimate OS authentication mechanisms, retrieval of encrypted secondary payloads from attacker-controlled infrastructure, execution from temporary files, and cleanup of artifacts such as terminal clearing and deletion of temporary payload files. Persistence mechanisms directly mentioned in the content include installation under hidden .npm_telemetry paths, shell startup file modifications (~/.zshrc, ~/.bashrc, ~/.bash_profile), and Linux @reboot cron entries. Exfiltration was reported to attacker-controlled servers, including infrastructure associated with trackpipe.dev; some reporting also mentions Telegram Bot API and GoFile.io as exfiltration channels.

Associated infrastructure and indicators directly mentioned in the content include trackpipe.dev, cloudcraftshub.com/api, dropras.xyz, IP 146.19.24.131, and malicious GitHub repositories such as Needvainverter93/deepseek-claw and others tied to the campaign. The malware has been associated in reporting with malicious repositories impersonating developer tools, trading bots, SDKs, and AI skills, and with campaigns targeting developers and AI-driven workflows across macOS, Linux, and Windows.

Share:
For your environment

Hunt this family in your stack

Mallory pivots from this family to the IOCs, detections, and named campaigns that touch your stack, and pages you when something new lands.

MITRE ATT&CK

Techniques & procedures

35 distinct techniques documented for this family, organized by ATT&CK tactic.

Reconnaissance

1 technique
T1598Phishing for InformationEvidence1

this script acts as a sophisticated dropper that uses terminal-based social engineering, such as spoofed sudo password prompts, to trick users into handing over credentials

Initial Access

5 techniques
T1078Valid AccountsEvidence1

Jamf Threat Labs exposes new GhostClaw/GhostLoader samples using malicious GitHub repos and AI dev workflows to steal macOS credentials via multi-stage payloads.

T1195Supply Chain CompromiseEvidence3

Zscaler’s analysts noted that as AI agents become standard in development pipelines, supply chain poisoning through fake skills is a growing threat. Their analysis revealed that the threat actor published the deceptive skill on GitHub, knowing that AI agents and developers would likely pull it into automated workflows without a second thought.

T1195.001Compromise Software Dependencies and Development ToolsEvidence1

"identified a live malicious npm package named @openclaw-ai/openclawai ... masquerades as a legitimate CLI tool"

T1566.002Spearphishing LinkEvidence1

the repositories contain a README with step-by-step installation instructions that encourage users to execute a shell command, typically using curl to retrieve and run a remote script.

T1566.003Spearphishing via ServiceEvidence1

A new malware campaign called GhostClaw is actively targeting macOS users through fake GitHub repositories and AI-assisted development workflows. The campaign uses social engineering disguised as legitimate developer tools to steal user credentials and drop secondary payloads on infected systems.

Execution

5 techniques
T1053.003CronEvidence1

"An @reboot cron entry is added"

T1059.004Unix ShellEvidence1

This retrieves and executes install.sh, which serves as the initial bootstrapper.

T1059.007JavaScriptEvidence5

The alternate attack path, built for macOS and Linux environments, used a heavily obfuscated Node.js file buried inside npm lifecycle scripts. When the install command ran, it silently dropped GhostLoader onto the system.

T1106Native APIEvidence1

"spawned as a detached child process" and "EXEC Execute arbitrary shell command"

T1204User ExecutionEvidence2

Execution begins when the user, or an automated agent, runs the provided installation command. This retrieves and executes install.sh, which serves as the initial bootstrapper.

Persistence

3 techniques
T1053.003CronEvidence1

"An @reboot cron entry is added"

T1078Valid AccountsEvidence1

Jamf Threat Labs exposes new GhostClaw/GhostLoader samples using malicious GitHub repos and AI dev workflows to steal macOS credentials via multi-stage payloads.

T1546.004Unix Shell Configuration ModificationEvidence1

"Lines are appended to ~/.zshrc, ~/.bashrc, and ~/.bash_profile"

Privilege Escalation

3 techniques
T1053.003CronEvidence1

"An @reboot cron entry is added"

T1078Valid AccountsEvidence1

Jamf Threat Labs exposes new GhostClaw/GhostLoader samples using malicious GitHub repos and AI dev workflows to steal macOS credentials via multi-stage payloads.

T1546.004Unix Shell Configuration ModificationEvidence1

"Lines are appended to ~/.zshrc, ~/.bashrc, and ~/.bash_profile"

Stealth

6 techniques
T1027Obfuscated Files or InformationEvidence3

For macOS and Linux systems, the campaign deploys an obfuscated Node.js payload that installs GhostLoader to steal sensitive developer data.

T1070Indicator RemovalEvidence1

"The NUKE command performs complete self-destruction... removes shell hooks... cleans cron jobs... deletes ... install directory"

T1070.003Clear Command HistoryEvidence1

enabling it steal system credentials, deliver the GhostLoader malware by contacting a command-and-control (C2) server, and remove traces of malicious activity by clearing the Terminal.

T1070.004File DeletionEvidence2

Following execution, the temporary file is removed... Following execution of the primary payload, postinstall.js is invoked to extend the compromise and obscure earlier activity.

T1078Valid AccountsEvidence1

Jamf Threat Labs exposes new GhostClaw/GhostLoader samples using malicious GitHub repos and AI dev workflows to steal macOS credentials via multi-stage payloads.

T1140Deobfuscate/Decode Files or InformationEvidence1

"payload is decrypted using AES-256-GCM"

Credential Access

4 techniques
T1056Input CaptureEvidence1

"displays a fake Keychain authorization prompt... victim is prompted for their system password (up to 5 attempts)"

T1539Steal Web Session CookieEvidence2

Remcos set itself to stealth mode immediately upon execution, logging keystrokes, stealing browser cookies... With Remcos giving attackers a full remote shell and GhostLoader scooping up cloud tokens, SSH keys, and browser session cookies...

T1552.004Private KeysEvidence2

Once active, GhostLoader swept through the host for anything valuable: macOS Keychain data, SSH keys, cryptocurrency wallet files, and cloud API tokens.

T1555Credentials from Password StoresEvidence1

Once active, GhostLoader swept through the host for anything valuable: macOS Keychain data, SSH keys, cryptocurrency wallet files, and cloud API tokens.

Discovery

4 techniques
T1016System Network Configuration DiscoveryEvidence1

"Geo Lookup hxxps://ipinfo.io/json"

T1082System Information DiscoveryEvidence1

"sends ... hostname, IP, country, and system specs" and "Geo Lookup hxxps://ipinfo.io/json"

T1083File and Directory DiscoveryEvidence1

"Files in ~/Desktop, ~/Documents, and ~/Downloads are scanned for BIP-39 mnemonic seed phrases"

T1526Cloud Service DiscoveryEvidence1

With Remcos giving attackers a full remote shell and GhostLoader scooping up cloud tokens, SSH keys, and browser session cookies... Once active, GhostLoader swept through the host for anything valuable... cloud API tokens.

Collection

3 techniques
T1056Input CaptureEvidence1

"displays a fake Keychain authorization prompt... victim is prompted for their system password (up to 5 attempts)"

T1115Clipboard DataEvidence1

"Clipboard monitor... Every 3 seconds, reads the system clipboard"

T1560Archive Collected DataEvidence1

"All collected data is compressed into a tar.gz archive"

Command and Control

5 techniques
T1071Application Layer ProtocolEvidence3

the next-stage downloader... reaches out to a Telegram channel to fetch the URL for the final payload... The initial npm package captures credentials and fetches configuration from either a Telegram channel or a Teletype.in page...

T1071.001Web ProtocolsEvidence1

"HTTP upload to hxxps://trackpipe.dev" and "polls the C2 panel every ~25 seconds"

T1090.001Internal ProxyEvidence1

"installs a persistent RAT... including a SOCKS5 proxy" and "PROXY_START Start a SOCKS5 proxy"

T1105Ingress Tool TransferEvidence2

Should they enter the password, the malware then silently retrieves the next-stage downloader, which then reaches out to a Telegram channel to fetch the URL for the final payload and the key required to decrypt it.

T1219Remote Access ToolsEvidence1

"persistent RAT with full remote access capabilities" and "polls the C2 panel ... supports EXEC"

Exfiltration

3 techniques
T1041Exfiltration Over C2 ChannelEvidence1

Once active, GhostLoader swept through the host for anything valuable: macOS Keychain data, SSH keys, cryptocurrency wallet files, and cloud API tokens. All of it was sent back to attacker-controlled servers.

T1567Exfiltration Over Web ServiceEvidence1

"GoFile.io ... upload.gofile.io/uploadfile with the hardcoded bearer token"

T1567.002Exfiltration to Cloud StorageEvidence1

"sends a 'New Session' notification to Telegram" and "Direct file upload via Bot API"

INDICATORS OF COMPROMISE

IOCs tracked for this family

32 indicators attributed across vendor reports, sandbox runs, and researcher write-ups. Full values are available in Mallory.

View more in app
Network
8 tracked

IPs, domains, and DNS infrastructure linked to this family.

Hashes
20 tracked

File hashes (MD5, SHA-1, SHA-256) from samples and reports.

Other
4 tracked

Other indicator types observed in public reporting.

TypeValueLatest sighting
domain●●●●●●●●●●●●View more in app3 days ago
domain●●●●●●●●●●●●View more in app5 days ago
domain●●●●●●●●●●●●View more in app21 days ago
domain●●●●●●●●●●●●View more in app2 months ago
uri●●●●●●●●●●●●View more in app2 months ago
domain●●●●●●●●●●●●View more in app2 months ago
ACTIVITY FEED

Recent activity

9 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.

gurucul threat researchNews
May 7, 2026
Malicious OpenClaw Skill Distributes Remcos RAT and GhostLoader | Community Portal | Gurucul

An obfuscated Node.js-delivered payload used on macOS and Linux to steal sensitive developer data.

Read more
cyber security newsNews
May 6, 2026
Malicious OpenClaw DeepSeek Skill Exploits Agentic AI Workflows to Deliver RAT and Stealer - Cyber Security News

A cross-platform stealer delivered through obfuscated Node.js code in npm lifecycle scripts from the fake DeepSeek-Claw skill. It targets developer environments, steals macOS Keychain data, SSH keys, cryptocurrency wallet files, cloud API tokens, and browser/session-related data, and exfiltrates them to attacker-controlled servers. On macOS and Linux it also presents fake password prompts to harvest credentials.

Read more
zscaler threat labzNews
May 5, 2026
OpenClaw Skill Distributes Remcos & GhostLoader | ThreatLabz

Cross-platform information stealer targeting developer environments. In this campaign it is delivered through manual installation paths such as install.sh or npm install, including an obfuscated Node.js payload. It harvests credentials via terminal-based social engineering and exfiltrates macOS keychain data, SSH keys, cryptocurrency wallets, and cloud API tokens.

Read more
the hacker newsNews
Mar 24, 2026
Ghost Campaign Uses 7 npm Packages to Steal Crypto Wallets and Credentials

A multi-stage macOS-focused malware used in the Ghost/GhostClaw campaign. It is delivered via malicious npm packages and GitHub repositories, tricks users into entering sudo credentials, contacts Telegram/C2 infrastructure to fetch later-stage payloads, steals credentials and sensitive data including cryptocurrency wallets, and can function as a remote access trojan awaiting further instructions.

Read more
What this page doesn’t show

The version that knows your environment.

This page is what’s public. Mallory adds the parts that aren’t: which of your assets match these IOCs, which detections are missing, which campaigns to expect next, and what to do in the next 30 minutes.
IOC matching32

Match every observed IP, domain, and hash against your live telemetry.

Threat actor attribution

Named campaigns wielding this family, with evidence pinned to each claim.

Exploited vulnerabilities

CVEs this family uses for access and lateral movement.

Detection signatures

YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.

MITRE ATT&CK mapping35

Every documented technique, ranked by evidence weight.

Researcher chatter

Reddit, Mastodon, and CTI community discussion around this family.