TrapDoor malware targets crypto developers and tools

TLDR

TrapDoor is an active malware campaign using poisoned open-source developer packages to steal crypto wallets, cloud keys, and GitHub tokens from crypto and AI developers.

1. TrapDoor spreads through malicious npm, PyPI, and Rust packages that auto-execute and exfiltrate secrets from developer machines.
2. It specifically targets crypto and DeFi tooling, including wallets and ecosystems like MetaMask, Solana, Sui, Aptos, Coinbase, Binance, and Brave.
3. Developers should urgently audit dependencies, rotate credentials, and review AI assistant configs, as this is a supply-chain threat to entire projects, not just individuals.

Deep Dive

1. How TrapDoor Works

Security firm Socket and others describe TrapDoor as an ongoing software supply chain attack distributed via more than 34 malicious packages and over 380 versions on npm, PyPI, and Crates.io. These packages masquerade as audit tools, build helpers, or security scanners for crypto, DeFi, AI, and infrastructure projects, then auto-execute via npm postinstall hooks, Python import side effects, or Rust build scripts to run on developer machines. After running, they scan for SSH keys, wallet keystores, API tokens, browser login databases, and environment variables, then send the data to attacker-controlled servers.

2. Who Is Targeted And What Data Is At Risk

Reports highlight a focus on developers building for Solana, Sui, Aptos, DeFi protocols, and AI systems, where stolen credentials give direct access to funds and critical infrastructure. TrapDoor seeks wallet-related data tied to services like Coinbase, Binance, MetaMask, and the Brave browser, as well as GitHub tokens and cloud credentials that can later be used to poison more repositories or reach production environments. This shifts the threat surface from on-chain smart contract bugs to developer workstations, which are often less tightly locked down than production infrastructure.

3. Why This Matters And What To Watch

Researchers note TrapDoor is low-volume but high-impact, prioritizing high-value developer environments over broad spraying. It also tries to hijack AI coding assistants such as Claude and Cursor by inserting hidden prompt instructions into files like CLAUDE.md or .cursorrules, tricking them into running workflows that discover and leak secrets. Practical responses at a high level include: auditing recent dependency additions and removing unknown security or helper packages, isolating and rebuilding compromised machines, and rotating all wallets, cloud keys, and GitHub tokens that may have touched those environments.

Confidence: high because multiple independent security firms have described consistent indicators of compromise and attack behavior.

Conclusion

TrapDoor is a sophisticated supply chain campaign aimed at the people and tools that build crypto, not just at end users. For anyone developing wallets, DeFi apps, or AI-integrated crypto tools, treating dependency hygiene, credential segregation, and AI assistant configuration as first-class security concerns is now critical to reducing the blast radius of similar attacks.