TLDR
A new malware campaign called TrapDoor is actively targeting crypto and AI developers through poisoned open source packages to steal wallets, keys, and cloud credentials.
- TrapDoor uses malicious npm, PyPI, and Rust (Crates.io) packages that masquerade as dev tools for ecosystems like Solana, Sui, and Aptos.
- Once installed, it tries to exfiltrate wallet keystores, SSH keys, AWS and GitHub tokens, and even hijack AI coding assistants such as Claude and Cursor.
- Developers who may have installed suspicious security or chain-specific helper packages should audit dependencies, rotate secrets, and harden their key storage practices.
Deep Dive
1. How TrapDoor Works
Security firm Socket reports that TrapDoor is an active software supply chain attack distributing more than 34 malicious packages and 384 versions across npm, PyPI, and Crates.io, disguised as common dev utilities and Sui/Move build helpers for crypto and AI projects. A detailed summary appears in a recent Cointelegraph report on TrapDoor.
These packages trigger via normal mechanisms such as npm postinstall scripts, Python imports, or Rust build.rs, so just adding them as dependencies can execute the payload.
SlowMist separately issued an emergency alert describing TrapDoor as a cross?registry campaign that fully compromises developer workstations, particularly in Solana, DeFi, and AI ecosystems, by embedding itself into project tooling and configuration files.
You can be compromised simply by installing what looks like a useful dev helper or security scanner, even if you never directly run a suspicious binary.
2. What It Steals And Who Is At Risk
According to Socket, TrapDoor scans infected machines for wallet data, SSH keys, cloud credentials (for example AWS), GitHub tokens, browser extension data, and API keys, then sends them to attacker servers, giving access to both crypto funds and production infrastructure.
The campaign is tuned to popular crypto wallets and environments, including MetaMask, Coinbase and Binance browser wallets, Braves wallet, and tooling around Solana, Sui, Aptos and other chains, as described in the Socket-based coverage.
U.Todays summary of a SlowMist bulletin adds that TrapDoor persists by modifying AI assistant config files like `.cursorrules` and `CLAUDE.md` and hiding in Git hooks and automation scripts, which can keep the infection alive across projects and vibe coding workflows.
Any developer workstation that touches real wallets, production keys, or privileged repos is a high?value target, not just crypto company servers.
3. Practical Risk Reduction Steps
Reports from Socket and SlowMist converge on three main defensive themes:
- Audit dependencies and remove unknown or newly added security, scanner, or chain?specific helper packages that are not strictly necessary.
- Rotate secrets aggressively if you suspect exposure: revoke and reissue wallet keys (where possible), API keys, cloud credentials, and GitHub tokens used on the affected machine.
- Move long?term holdings to hardware wallets or other isolated storage, and keep only minimal hot?wallet balances on machines used for development.
For AI?assisted workflows, review and clean configuration files for tools like Claude and Cursor, and treat any auto?generated security scan or key discovery workflow with skepticism.
The safest posture is to assume that a compromised dev box reveals everything it touches, then shrink what that box can see and rapidly replace any credentials it might have leaked.
Conclusion
TrapDoor highlights that the weakest link in crypto security can be a developers everyday tooling rather than a smart contract or exchange. By poisoning open source packages and hijacking AI assistants, attackers aim straight at wallets, keys, and infrastructure behind crypto projects. The key response is to harden development environments, minimize on?device wallet exposure, and treat dependency hygiene and secret rotation as core parts of crypto security, not optional extras.
CoinsKid.Com