Internet Bug Bounty · Report #2293731
Command Injection using malicious hostname in expanded proxycommand
Summary
Subject: Security weakness in ProxyCommand handling CVE ID#: CVE-2023-6004 Versions: 0.10.x, 0.9.x, 0.8.x Summary: ProxyCommand/ProxyJump features enable to inject malicious code through hostname
Description Using the ProxyCommand or the ProxyJump feature enables users to exploit unchecked hostname syntax on the client, which enables to inject malicious code into the command of the above-mentioned features through the hostname parameter.
User interaction is required to exploit this issue.
Credits Originally reported by Vinci (https:// github.com/vin01). Patches provided by Norbert Pocs of the libssh team.
Full Security Advisory: https://www.libssh.org/security/advisories/CVE-2023-6004.txt
How it was found
Using the ProxyCommand or the ProxyJump feature enables users to exploit unchecked hostname syntax on the client, which enables to inject malicious code into the command of the above-mentioned features through the hostname parameter.
User interaction is required to exploit this issue.
Advisory from libssh: https://www.libssh.org/security/advisories/CVE-2023-6004.txt
Advisory from OpenSSH which also suffered from this flaw: https://www.openssh.com/txt/release-9.6
Impact
Code execution via malicious input hostname or other tokens