Trigram Campaign Masquerades as 50 AI & Blockchain Brands to Steal Devs’ Data

A new industrial-scale intrusion initiative dubbed “Trigram” is targeting developers in the U.S. and Europe by impersonating around 50 leading AI and blockchain brands. The campaign’s aim: tricking devs into running infostealers using multiple access routes, including phishing, fake software packages, and supply-chain tricks.

How Trigram Operates

The campaign relies on impersonation of companies operating in the AI, blockchain, education, logistics, and technology spaces. Attackers flex social engineering techniques—involving fake job offers, conference invitations, and package delivery notices—to lure developers into taking actions that open the door for malware deployment. A major part of the attack chain involves a trojanized repository, compromised software supply chains, and deceptive emails bearing recruiter or conference themes. All paths lead to the same backend infrastructure and drop the same infostealing payloads. (Stormshield in its recent investigation provides the technical breakdown.)

One method involves compromised code in npm packages: a fake “antd-img-crop” commit installed through a “prebuild hook” that triggers malware when developers run “npm run build.” The infection sequence then pulls in a multi-stage loader, the node.js environment if missing, and ultimately downloads a JavaScript file (“install_guard.js”) that harvests credentials, browser data, wallets, SSH keys, and environment files. Similarly, on macOS, targets have been shown a fake dialog to grant system privileges under the guise of verifying drivers or security settings before the payload activates.

Links to Kimsuky and UNK_DeadDrop

Security researchers believe Trigram is closely related to UNK_DeadDrop, a North Korea-aligned threat group. It shares infrastructure, destructive tools, and tactics with that cluster and with Contagious Interview and Operation GitPower. For example, Trigram uses cloned identities, email infrastructure overlaps, fake recruiter personas, and domain names tied to previous campaigns. One fake recruiter impersonates Amber Group, and victims receive bogus job or invite-to-conference emails. In other instances, a courier-style website or malicious Visual Studio Code/extension impersonates “google-dev-tools.google-update-support.”

Kimsuky also appears tied via Operation GitPower, targeting experts in Ukraine across diplomacy, academia, policy, and security research. Attack vectors there include flagged LNK file attachments passed off as conference invites and fake Zoom installers that hide malware behind legitimate applications.

Infrastructure, Tools & Victim Profiling

Underlying the campaign is a complex infrastructure that includes:

  • Email delivery through multiple legitimate services;
  • Use of GitHub and GitLab as dead drops and configuration hosts;
  • A tracking and management backend (dubbed “rms-platform”) that rates visitors, applies geofencing, and filters for candidates using invitation codes.

Visitors first land on fake portals mimicking recruiters or shipping companies. If they pass geofence filters and have a valid “invitation code,” they’re guided into executing commands that fetch malicious scripts. Once the system detects a suitable target, the code delivered depends on the operating system—Windows, Linux or macOS.

When the attack hits its final phase, it sets up an info-stealer that copies passwords, cookies, wallets, SSH keys, environment variables, and other sensitive data. It even impersonates Google software during remote-access agent installation.

Why This Matters

Trigram significantly broadens both the scale and methods seen in prior North Korea-linked hacking campaigns. Instead of one delivery vector, this campaign uses several angles—job fraud, code repository poisoning, fake tools—all aligned toward compromising developers. It also boosts automation, using real-time victim tracking and role-based candidate filters.

This shift raises the stakes. Developers now face a multi-front risk—social engineering, supply chain, malicious open-source components—often delivered via familiar services and tools. Awareness of the threat model, carefully gating of code and dependencies, and strong verification of recruiter or conference sources are more essential than ever.