A malware campaign targets developers using AI coding assistants by compromising GitHub repositories. Attackers exploit AI agent configuration files like AGENTS.md and CLAUDE.md to deliver a concealed payload (setup.js) that steals credentials from multiple cloud and code platforms. The malware validates stolen tokens, modifies repositories, infects software packages, and propagates by injecting malicious instructions into AI-agent configuration files. It spreads further by committing malicious files through victim accounts with sufficient permissions, impacting major package ecosystems such as npm, PyPI, RubyGems, and JFrog.
An investigation reveals that placeholder domains commonly used in documentation are serving malicious content through advertising chains. Two domains, yoursite[.]com and your-domain[.]com, present in hundreds of thousands of GitHub files and agent skills, use cloaking techniques to display scams specifically to macOS visitors while showing legitimate parking pages to others. The scams include fake MacOS Security Center warnings and counterfeit news articles promoting investment schemes. These attacks leverage affiliate fraud models, funneling victims through multiple redirectors to genuine services like McAfee subscriptions to earn commissions. Static security checks fail to detect these threats because malicious redirects occur after JavaScript execution. This follows an earlier disclosure about third-party[.]com serving ClickFix lures. The widespread citation of unregistered placeholder domains in documentation creates significant exposure, particularly for AI agents that fetch and act on this content.
The Intercom TypeScript Library version 7.0.4 has been compromised with malicious code that harvests GitHub credentials. Upon installation, the package executes a preinstall hook that downloads the Bun runtime, then runs a payload to extract GitHub credentials using the gh auth token command. The attack employs sophisticated C2 communication by querying GitHub's commit search API for specific strings embedded in public repositories, effectively using legitimate services to evade detection. The attack patterns mirror previous Shai-Hulud compromises, which exhibit worm-like behavior by automatically using stolen credentials to infect additional npm packages. With 361,510 weekly downloads, this compromise poses significant risk for a widespread infection wave similar to November 2025 when over 1,000 packages were affected.
A large-scale malware campaign tracked as TroyDen's Lure Factory has been identified distributing LuaJIT-based infostealers through over 300 delivery packages hosted on GitHub. The operation uses AI-generated lure names incorporating obscure biological taxonomy and medical terminology to target developers, gamers, Roblox players, and crypto users. The malware employs a two-component design with a renamed LuaJIT runtime and encrypted Lua payload that evades sandbox detection through anti-analysis checks and extreme sleep delays. Upon execution, it disables proxy detection, captures desktop screenshots, performs geolocation, and exfiltrates data to C2 servers in Frankfurt. The infrastructure demonstrates scalability with multiple IP addresses serving identical encrypted commands, while maintaining simultaneous campaigns across gaming cheats, developer tools, phone trackers, and VPN crackers.
The npm ecosystem experienced a critical shift in September 2025 with the Shai-Hulud worm, marking the transition from isolated attacks to systematic supply chain compromises. In April 2026, TeamPCP launched a coordinated campaign through a malicious @bitwarden/cli package targeting multiple distribution channels including Docker Hub, GitHub Actions, and VS Code extensions. The multi-stage payload employs advanced obfuscation, harvests credentials from cloud providers and developer workstations, exfiltrates data through encrypted HTTPS and GitHub repositories, and self-propagates by backdooring npm packages using stolen tokens. The malware implements GitHub's search API as a resilient command-and-control fallback mechanism and features anti-detection measures including Russian locale killswitches. This represents an evolution toward wormable propagation, infrastructure-level persistence, and dormant payloads that activate under specific conditions.
The GhostClaw malware campaign has expanded its distribution methods beyond npm packages to include GitHub repositories and AI-assisted development workflows. The attackers impersonate legitimate tools and utilize multi-stage payloads to steal credentials and retrieve additional malicious code. The infection chain involves executing shell commands, presenting fake authentication prompts, and establishing persistence. The campaign leverages both manual installation through README instructions and automated AI-assisted workflows. Multiple GitHub repositories have been identified, all communicating with a common command-and-control infrastructure. This shift in tactics allows the attackers to target a broader range of victims, including developers and users of AI-assisted coding tools.
The Contagious Trader campaign is a sophisticated malware operation targeting cryptocurrency users, attributed to North Korea with high confidence. It involves malicious cryptocurrency trading bot projects on GitHub that exfiltrate sensitive data and private keys using various techniques, including malicious npm dependencies. The campaign demonstrates overlaps with known North Korean tactics, particularly those of FAMOUS CHOLLIMA, including the use of GitHub, npm, and Vercel infrastructure, Base64-encoded payload URLs, and anonymizing VPNs for npm package publishing. The operation represents a shift in tactics, expanding beyond the previous Contagious Interview campaign to target a broader range of cryptocurrency users.
An active supply chain worm campaign, dubbed SANDWORM_MODE, is spreading through typosquatting and AI toolchain poisoning across at least 19 malicious npm packages. The worm exhibits Shai-Hulud characteristics, incorporating GitHub API exfiltration with DNS fallback, hook-based persistence, SSH propagation, and MCP server injection targeting AI coding assistants. It harvests credentials from developer and CI environments, exfiltrates data via multiple channels, and uses stolen identities to propagate. The campaign also includes a weaponized GitHub Action for CI secret harvesting. The worm employs a multi-stage design with obfuscated loaders, time-gated execution, and extensive configuration options. It targets high-traffic developer utilities, crypto tooling, and AI coding tools, posing a significant threat to the software supply chain.
A malicious fork of the MacOS app Triton was discovered on GitHub, containing Windows-targeted malware disguised as the legitimate application. The attacker modified the repository, redirecting download links to a ZIP file hosting the malware. Analysis revealed sophisticated evasion techniques, anti-analysis features, and potential cryptocurrency functionality. The low detection rate and peculiar implementation suggest either an amateur attempt or a possible AI-generated attack. The incident highlights broader concerns about GitHub's security practices and Microsoft's priorities, prompting a call for developers to consider alternative code hosting platforms that better align with open-source values and user privacy.
RedKitten is a newly identified campaign targeting Iranian interests, first observed in January 2026. The malware uses GitHub and Google Drive for configuration and payload retrieval, and Telegram for command and control. It appears to exploit the Dey 1404 Protests in Iran, targeting organizations documenting human rights abuses. The threat actor rapidly built this campaign using AI tools, as evidenced by traces of LLM-assisted development. While attribution is not definitive, the activity aligns with Iranian state-sponsored attackers. The malware, dubbed SloppyMIO, can fetch modules, execute commands, collect files, and deploy additional malware with persistence.
A Pakistan-linked APT group conducted two campaigns targeting Indian government entities. The Gopher Strike campaign used PDFs with malicious links to deliver an ISO file containing GOGITTER, a Golang downloader that fetches payloads from private GitHub repositories. GITSHELLPAD, a Golang backdoor, was used for C2 communication via GitHub. GOSHELL, a Golang shellcode loader, deployed Cobalt Strike Beacon on specific hostnames. The attackers used various techniques including scheduled tasks for persistence, obfuscation, and environmental keying. Post-compromise activities involved system reconnaissance and data exfiltration. The campaign demonstrated sophisticated TTPs and custom-built tools, indicating a potentially new subgroup or parallel Pakistan-linked threat actor.
In January 2026, IIJ observed malicious LNK files targeting Korean users to execute the MoonPeak malware, attributed to North Korean threat actors. The infection chain begins with a LNK file that runs an obfuscated PowerShell script, which checks for analysis environments, creates additional scripts, and sets up persistence. The second stage downloads and executes a payload from GitHub, which is actually the MoonPeak malware. MoonPeak is obfuscated using ConfuserEx and communicates with a C2 server. The campaign utilizes GitHub for hosting malware, a technique known as Living Off Trusted Sites (LOTS). This attack demonstrates the ongoing threat posed by North Korean actors targeting various countries and individuals worldwide.
An active infostealer campaign is targeting macOS and Windows users across various sectors. The threat actors are using SEO poisoning to direct victims to fake GitHub repositories impersonating legitimate tools like PagerDuty. The campaign involves over 20 malicious repositories active since September 2025. The attack flow begins with a Google search, leading to a fraudulent GitHub repository, then to a GitHub Pages site with a deceptive command. This command deploys the MacSync stealer in three stages: a loader, a dropper, and the final payload. MacSync aggressively harvests credentials from browsers, cloud services, and cryptocurrency wallets. The campaign's scale includes 39 identified malicious repositories, with 24 still active as of January 2026. Evasion tactics include using 'readme-only' repositories and distributed identities.
The Contagious Interview campaign, attributed to North Korea, continues to target software developers through fake recruitment schemes. A new technique in their arsenal leverages Microsoft Visual Studio Code task files to execute malicious code when a project is opened. The report documents observations of this vector, presents GitHub-based discovery methods, highlights findings including a new malicious NPM package, and outlines detection opportunities. The campaign exploits VS Code's Task feature, using the runOptions property to automatically execute malicious shell commands when a workspace is opened. Various obfuscation techniques are employed, including hiding commands with whitespace and masquerading payloads as image or font files.
PurpleBravo, a North Korean state-sponsored threat group, targets software developers through fake recruitment efforts, particularly in cryptocurrency and software development sectors. Their toolkit includes BeaverTail, PyLangGhost, and GolangGhost, designed for stealing browser credentials and cryptocurrency information. The group has affected 3,136 IP addresses, mainly in South Asia and North America, compromising 20 organizations across various industries. PurpleBravo's tactics include using fictitious personas, malicious GitHub repositories, and sophisticated malware to infiltrate IT services companies, posing a significant supply-chain risk. The group shows overlap with PurpleDelta, another North Korean threat actor, sharing infrastructure and operational patterns. PurpleBravo's focus on the IT sector in South Asia presents an overlooked threat to organizations outsourcing IT services.
North Korean threat actors have evolved their techniques in the Contagious Interview campaign, now abusing Microsoft Visual Studio Code task configuration files. The infection chain begins when a victim opens a malicious Git repository, often disguised as part of a recruitment process. If trust is granted, arbitrary commands are executed on the system. The malware uses JavaScript payloads hosted on vercel.app to implement backdoor logic, including remote code execution, system fingerprinting, and persistent command-and-control communication. The backdoor collects host information and beacons to a C2 server every five seconds. Recent observations show further execution of similar payloads, indicating ongoing development of these tactics.
Shai-Hulud V2 is an advanced malware campaign targeting the npm software supply chain, compromising over 700 npm packages and creating more than 27,000 malicious GitHub repositories. It introduces sophisticated techniques including pre-install phase execution, persistent backdoors via self-hosted GitHub Actions runners, credential harvesting and recycling across victims, and a destructive failsafe mechanism. The malware exfiltrates data through GitHub and propagates within the npm ecosystem, also exploiting Azure DevOps build agents. This supply chain attack enables persistent remote code execution and widespread credential theft without requiring user interaction. European organizations relying on npm packages and GitHub Actions for CI/CD pipelines face significant risks of data breaches, system compromise, and operational disruption. Mitigation requires enhanced supply chain security practices, strict GitHub Actions runner controls, credential hygiene, and proactive monitoring for anomalous activity. Countries with strong software development sectors and high npm usage, such as Germany, France, the UK, and the Netherlands, are particularly vulnerable. Given its broad impact and advanced persistence mechanisms, the threat severity is assessed as high.
The Shai-hulud 2.0 campaign is an advanced malware operation targeting cloud platforms (AWS, GCP, Azure) and developer ecosystems by stealing credentials and secrets. It automates backdooring of NPM packages maintained by victims, enabling rapid supply chain propagation. The malware abuses GitHub Actions workflows for command-and-control and secret exfiltration, leveraging cloud secret management services and destructive failsafes to maintain stealth and persistence. This campaign threatens the confidentiality and integrity of cloud and developer environments, potentially impacting thousands of downstream users. It requires no known exploits in the wild but uses sophisticated tactics to evade detection. European organizations relying heavily on cloud services and open-source development are at risk, especially those with significant use of NPM packages and GitHub repositories. The threat severity is assessed as high due to the broad impact on supply chains and credential theft without requiring user interaction. Immediate mitigation includes securing cloud credentials, auditing GitHub workflows, and monitoring supply chain dependencies for unauthorized changes.
Shai-Hulud 2.0 is a highly aggressive and automated supply-chain malware targeting the npm ecosystem, identified in November 2025. It rapidly compromises hundreds of npm packages within hours, behaving like a worm that harvests credentials and cloud secrets. The malware leverages GitHub Actions as a persistent backdoor and creates public repositories to exfiltrate stolen data. This attack represents a significant escalation in supply-chain attack sophistication, affecting major projects and organizations globally. It results in tens of thousands of attacker-created GitHub repositories, facilitating widespread propagation. The malware automates spreading to new npm accounts, increasing infection speed and scale. It does not require user interaction once initial compromise occurs and exploits trusted software supply chains. No CVE or patch is currently available, and no known exploits in the wild have been reported yet. The attack’s medium severity rating may underestimate its potential impact given its worm-like behavior and credential theft capabilities.
APT-C-60 continues to target Japan and East Asia with spear-phishing attacks impersonating job seekers. The attack flow has evolved, now directly attaching malicious VHDX files to emails. The malware, including Downloader1, Downloader2, and SpyGlace, has been updated with new features and communication methods. SpyGlace versions 3.1.12, 3.1.13, and 3.1.14 were observed, with changes in Mutex values and execution paths. The attackers use GitHub for payload distribution and employ sophisticated encoding and encryption techniques. The campaign abuses legitimate services and maintains consistent behavioral patterns despite infrastructure changes.