Headline
CVE-2023-34254: Remoteinventory task command injection when using ssh command mode
The GLPI Agent is a generic management agent. Prior to version 1.5, if glpi-agent is running remoteinventory task against an Unix platform with ssh command, an administrator user on the remote can manage to inject a command in a specific workflow the agent would run with the privileges it uses. In the case, the agent is running with administration privileges, a malicious user could gain high privileges on the computer glpi-agent is running on. A malicious user could also disclose all remote accesses the agent is configured with for remoteinventory task. This vulnerability has been patched in glpi-agent 1.5.
Impact
If GLPI-Agent is running remoteinventory task against an Unix platform with ssh command, an administrator user on the remote can manage to inject a command in a specific workflow the agent would run with the privileges it uses. In the case, the agent is running with administration privileges, a malicious user could gain high privileges on the computer glpi-agent is running on. A malicious user could also disclose all remote accesses the agent is configured with for remoteinventory task.
Patches
Upgrade to GLPI-Agent 1.5
Workarounds
Upgrade to GLPI-Agent latest nightly build and force remoteinventory mode to libssh2 adding ?mode=libssh2 to the remote url.
Or you can disable the vulnerable code by setting no-category configuration to process,virtualmachine,video,drive,database,storage.
For more information
If you have any questions or comments about this advisory, mail us at [email protected].