Security researchers have uncovered a large-scale software supply chain campaign named TrapDoor, which targeted developers through 34 malicious packages and more than 384 related versions distributed across npm, PyPI, and Crates.io. The operation primarily focused on cryptocurrency, DeFi, Solana, AI, and blockchain developers by disguising malware as legitimate development utilities and security auditing tools. The earliest known package, *eth-security-auditor@0.1.0*, appeared on PyPI on May 22, 2026, before the attackers rapidly expanded into multiple repositories using convincing package names such as *prompt-engineering-toolkit*, *solidity-deploy-guard*, and *defi-threat-scanner*. Security firm Socket identified the malicious activity within minutes of publication, helping reduce the risk of widespread compromise. The attackers customized their techniques for each ecosystem to ensure malicious code executed automatically during normal developer workflows. In npm, the malware abused post-installation hooks to deploy a persistent payload known as *trap-core.js*, which harvested credentials and sensitive information. PyPI packages executed code automatically when imported, downloading remote payloads hosted on GitHub Pages to dynamically update malicious behavior without releasing new package versions. Meanwhile, Crates.io packages relied on malicious Rust *build.rs* scripts that specifically searched for Sui and Move developer keystores. The malware targeted SSH keys, browser data, AWS credentials, and crypto wallets linked to Solana, Sui, and Aptos. To avoid detection, the campaign used multiple encryption methods including Fernet, ECDH, and XOR-based obfuscation. A particularly dangerous aspect of the TrapDoor operation was its abuse of AI coding assistants. Attackers modified *.cursorrules* and *CLAUDE.md* files using hidden zero-width Unicode characters to manipulate AI tools into secretly performing credential theft while appearing to run harmless security scans. Using the GitHub account *ddjidd564*, the threat actor submitted deceptive pull requests to well-known AI-related open-source projects including LangChain, MetaGPT, and OpenHands. The campaign also maintained an advanced command-and-control infrastructure through GitHub Pages, supported by a framework document named *AUDIT-MATRIX.md*, which described methods for disguising credential theft as legitimate developer automation tasks.
Cybersecurity researchers have identified an ongoing malware campaign distributing the VIP Keylogger malware through phishing emails and malicious attachments. The campaign primari...
ClearFake malware operators have introduced a new tactic by abusing Binance Smart Chain (BSC) Testnet infrastructure to host and retrieve malicious content, further advancing block...
Roundcube Webmail administrators are being advised to urgently patch their systems after the disclosure of several high-risk security vulnerabilities affecting both the 1.6.x and 1...