Scanning…
Caught in the wild
Real malicious packages from PyPI, run in the trap. Each one tried to send data to its owner's server; the trap caught it, and nothing reached the attacker.
Paste a GitHub link. We run the code in a locked sandbox with fake keys, let an AI agent work on it, and catch anything that tries to steal — down to the exact line.
Real malicious packages from PyPI, run in the trap. Each one tried to send data to its owner's server; the trap caught it, and nothing reached the attacker.