Mastra AI npm Packages Backdoored via easy-day-js Typosquat
A North Korean state-sponsored group hijacked a dormant npm account to inject a malicious typosquat dependency into 144 Mastra AI agent framework packages.
Microsoft Threat Intelligence has attributed a supply chain attack on the Mastra AI framework to the North Korean state-sponsored group Sapphire Sleet. The June 17 incident compromised the @mastra npm scope, exposing developers to credential theft and cryptocurrency-stealing malware. For teams building with the Mastra framework, which sees over 1.1 million weekly downloads, the breach requires an immediate audit of CI/CD environments.
The npm Hijack
Attackers targeted a dormant npm contributor account belonging to a former Mastra developer, username ehindero. Over an 84-minute window early on June 17, 2026, the attackers republished 144 packages under the @mastra scope. Affected packages included core framework components like mastra, @mastra/core, and create-mastra.
The attackers did not alter the framework’s source code. They injected a single malicious dependency named easy-day-js, a typosquat of the popular dayjs library.
Dropper Mechanics and Payload
The injected easy-day-js package operated as a multi-stage malware dropper triggered during the npm installation process.
When a developer or build pipeline installed a compromised Mastra package, easy-day-js executed a postinstall script. The script immediately disabled Transport Layer Security (TLS) certificate verification to bypass local network security monitoring. It then established a connection with an attacker-controlled command-and-control server.
The final payload was a cross-platform Remote Access Trojan (RAT) spawned as a detached, hidden process. Once running, the script deleted its own dropper files to hinder forensic analysis. The RAT actively hunted for authentication tokens, API keys, and cryptocurrency wallets.
Threat Actor Profile
Microsoft identified the attackers as Sapphire Sleet, also tracked as BlueNoroff, with high confidence on June 19. The group is a subset of North Korea’s state-sponsored cyber apparatus, traditionally focused on cryptocurrency theft to fund state operations.
The group’s tradecraft here matches their historical patterns. The use of clean-then-armed typosquats and detached spawning mechanisms aligns with earlier 2026 campaigns. Security firms including SafeDep, Snyk, and Orca Security detected the anomalous republishing events shortly after the 84-minute burst.
| Attack Detail | Information |
|---|---|
| Date of Compromise | June 17, 2026 |
| Attribution Date | June 19, 2026 |
| Affected Scope | 143-144 packages (@mastra) |
| Malicious Package | easy-day-js |
| Exploited Account | ehindero |
| Target Ecosystem | ~1.1 million weekly downloads |
Supply Chain Realities for AI Frameworks
Mastra AI reached version 1.0 in January 2026 and recently closed a $22M Series A. As an open-source TypeScript framework for agentic systems, it sits in highly privileged environments with access to production cloud keys and enterprise data stores.
This incident highlights that AI tooling remains highly vulnerable to conventional infrastructure exploits. The breach did not leverage prompt injection or model manipulation. Much like the recent LiteLLM PyPI package compromised by a supply chain attack, this event exploited structural weaknesses in package registries. The attackers relied entirely on the lack of mandatory credential expiration for inactive maintainers.
If your infrastructure pulled Mastra packages between June 17 and June 19, you must assume your environment is compromised. Rotate all cloud credentials, revoke exposed API keys, and audit your package lockfiles for the easy-day-js dependency.
Get Insanely Good at AI
The book for developers who want to understand how AI actually works. LLMs, prompt engineering, RAG, AI agents, and production systems.
Keep Reading
How to Use Symbolic Execution for Automated BPF Analysis
Learn how Cloudflare uses the Z3 theorem prover to instantly generate magic packets and reverse-engineer BPF bytecode for security research.
PyPI and npm Purge 73 Fake Azure Packages Targeting AI Agents
Security researchers discovered 73 malicious PyPI and npm packages mimicking Microsoft Azure libraries to install credential stealers on AI coding agents.
AI Prompt Injection Masks Malware in 19 PyPI Science Packages
The Hades supply chain campaign compromised 19 bioinformatics and Graph ML libraries on PyPI with memory scrapers and AI scanner misdirection.
OpenAI Secures ChatGPT macOS App After Axios Library Attack
OpenAI rotated its macOS code-signing certificates and hardened GitHub workflows following a dependency confusion attack on the ChatGPT desktop client.
Cisco Source Code and AWS Keys Stolen in Trivy Supply Chain Attack
Cisco confirms a major data breach involving stolen AI product source code and AWS keys following a malicious compromise of the Trivy vulnerability scanner.