TrapDoor malware targets crypto wallets and infrastructure

TLDR

TrapDoor is an active malware campaign that poisons developer tools to steal crypto wallets and critical infrastructure credentials.

1. TrapDoor hides in npm, PyPI, and Rust packages, targeting crypto, DeFi, AI, and security developers through their normal dev tooling.
2. The malware hunts for wallet keystores, SSH keys, AWS and GitHub credentials, threatening both user funds and backend crypto infrastructure.
3. Developers and teams should assume supply chain risk, audit dependencies, rotate secrets, and lock down AI coding assistants and build pipelines.

Deep Dive

1. What TrapDoor Is And How It Spreads

Security firm Socket and others report a coordinated supply chain attack called TrapDoor, with more than 34 malicious packages and around 384 versions across npm, PyPI, and Crates.io developer registries. Packages mimic legitimate tools such as security scanners, Solidity utilities, and Sui/Move build helpers, so they are likely to be installed as routine dependencies in crypto and AI projects.

Once installed, TrapDoor executes via ecosystem-specific hooks, such as npm postinstall scripts, Python import triggers, or Rust build scripts, blending into normal build and install workflows. Reports from Socket and Cointelegraph describe it as an ongoing low-volume but high-impact campaign that keeps releasing new versions to stay ahead of simple blocklists.

2. What TrapDoor Steals From Crypto Environments

TrapDoor is designed to exfiltrate high-value secrets rather than encrypt files. Analyses from Socket and multiple outlets say it targets wallet data, SSH keys, browser extension data, AWS cloud credentials, GitHub tokens, and other API keys, specifically calling out integrations with Coinbase, Binance, Solana, Sui, Aptos, MetaMask, and Brave wallet functions.

Because it can grab both local wallet keystores and infrastructure credentials, a single infected developer workstation can expose personal funds, private repos, deployment keys, and production systems. SlowMist notes that this mirrors prior Web3-focused worms, but with more emphasis on developer environments rather than end-user devices.

3. AI Tools, Persistence, And How To Mitigate

A distinctive angle is TrapDoors use of AI coding assistants. Malicious packages can write into configuration files like `.cursorrules` and `CLAUDE.md`, or Git hooks, to influence assistants such as Claude and Cursor into running fake security scans that actually discover and send secrets. Reports from crypto.news and U.Today show this as part of a broader shift toward abusing AI-assisted workflows.

Mitigations recommended by Socket and other researchers include: manually removing identified malicious packages, rebuilding compromised environments from clean images, rotating all credentials (wallet keys, SSH, cloud, GitHub), enforcing hardware wallets for significant funds, using lockfiles and pinning versions, and adding automated supply chain and behavior-based scanning in CI. Teams should also review AI assistant configuration and disable or harden any automation that can read secrets or run shell commands.

Conclusion

TrapDoor demonstrates how modern crypto attacks increasingly target developers and infrastructure rather than only end-user wallets, using trusted package registries and AI tooling as the entry point. For anyone building or maintaining crypto systems, tightening supply chain security, limiting where secrets live, and hardening AI-assisted workflows are now essential parts of protecting both user assets and critical backend infrastructure.