CVE-2023-28531 Vulnerability

The vulnerability lies within the ssh-add tool, which is used to load private keys into the ssh-agent authentication agent.

In OpenSSH 8.9, a feature was added to support "per-hop destination constraints" for keys stored on smartcards. This feature allows a user to load a key but restrict its usage to specific hosts (e.g., "allow this key only for ssh jump-host").

However, due to a logic flaw in the implementation, when a user executes the command to add a smartcard key with these constraints (e.g., ssh-add -h "constraint" ...), the constraints are never actually sent to the ssh-agent. The agent simply receives the key and stores it as an unrestricted key. This means the key can be used to authenticate to any server that accepts it, rather than just the intended targets.

Tactics, Techniques & Procedures (TTPs)

Attackers exploit CVE-2023-28531 primarily for Lateral Movement and Defense Evasion.

• Lateral Movement: If an attacker compromises a machine where a user has forwarded their SSH agent (believing it was restricted), the attacker can hijack that agent to authenticate to other servers on the network that the user has access to, bypassing the intended destination limits.

• Policy Bypass: The vulnerability inherently defeats security policies designed to compartmentalize access. Attackers leverage this to pivot through networks more freely than the administrator intended.

Indicators of compromise

Detecting active exploitation is difficult because the traffic appears as legitimate SSH authentication traffic. However, auditors can look for:

• Configuration Audits: Scanning for ssh-add commands in shell history that utilize the -h flag in conjunction with smartcard libraries.

• Agent Inspection: Users can run ssh-add -L to list keys currently in their agent. If a key was added with constraints but appears in the list without them, the vulnerability has triggered.

Known proof-of-concepts & exploits

There are no complex exploit scripts required for this vulnerability. The "exploit" is simply the standard usage of the tool in a vulnerable configuration.

Security researchers and vendors have confirmed that simply running the standard ssh-add command with constraints on a vulnerable version results in the failure, verifying the issue is widespread across unpatched Linux distributions.