I reproduced a Claude Code RCE. The bug pattern is everywhere.

A security researcher and the article author reproduced an RCE in Claude Code 2.1.118.

The flaw stems from a parsing anti-pattern that can let attacker-controlled input trigger code execution.

The author says this same mistake appears across many AI developer tools, not just Claude Code.

That means patching one bug may not be enough if the broader parsing pattern remains.