Authentication

The choice of the right authentication method against SSH-MITM can have a decisive influence on the success of a Man in the Middle attack.

The most popular authentication methods are “password” and “publickey” authentication. However, there are others, such as “none” and “keyboard-interactive”.

To log in to an SSH server it is necessary to specify an existing user. There are systems that use a default username. This is most common with Git repositories. Examples of this are GitLab and GitHub, which use the username “git” and distinguish which resources a user is allowed to access based on the public key at login.

none authentication - RFC-4252/5.2

The none authentication takes a special position among the authentication methods. The reason is that this authentication method is used to tell the client which methods are accepted by the server. For this reason, none-Authentication is executed before all other authentication methods.

However, it can also be used to give a user access to a system without requiring an explicit login.

5.2.  The "none" Authentication Request

A client may request a list of authentication 'method name' values
that may continue by using the "none" authentication 'method name'.

If no authentication is needed for the user, the server MUST return
SSH_MSG_USERAUTH_SUCCESS.  Otherwise, the server MUST return
SSH_MSG_USERAUTH_FAILURE and MAY return with it a list of methods
that may continue in its 'authentications that can continue' value.

This 'method name' MUST NOT be listed as supported by the server.

Support in SSH-MITM

none authentication is fully supported but disabled by default. The reason is, that this authentication method can break ssh-mitm attacks, if the remote server does not allow logins with none authentication

$ ssh-mitm server --remote-host 192.168.0.x --enable-none-auth

publickey authentication - RFC-4252/7

In contrast to password authentication, where the password is transmitted in plain text, publickey authentication is based on asymmetric encryption. In asymmetric encryption, a key pair consisting of a private part and a public part is created.

With SSH, the public key is stored on the server while the private key remains in the possession of the user.

With OpenSSH, the private key can be additionally protected by a password in case it is stolen. Starting with OpenSSH 8.2, FIDO2 tokens are supported and can be used for a 2 factor confirmation when using the key.

As an alternative to a FIDO2 token, SSH-Askpass can also be used to confirm cryptographic operations. If there is an increased need for security, SSH-Askpass should only be used if the use of a FIDO2 token is not possible.

Publickey authentication is only partially suitable for a man-in-the-middle attack. Unlike password authentication, no data is exchanged that can be used for a complete login.

Another problem is that the key that should be used for the login is not known. This can lead to FIDO2 and SSH-Askpass protected keys displaying a prompt for an incorrect key to the user.

An attacker should use PublicKey authentication only if the client does not accept other authentication methods.

Support in SSH-MITM

publickey authentication is supported and SSH-MITM is able to detect, if a user is able to login with publickey authentication on the remote server. This allows SSH-MITM to accept the same key as the destination server.

SSH-MITM is able to request the agent from the client and use it for remote authentication. By using this feature, it’s possible to do a full man-in-the-middle attack when publickey authentication is used.

If publickey authentication is not possible, the authentication will fall back to password-authentication.

Publickey authentication in SSH-MITM is enabled by default. All you have to do is to start the server:

$ ssh-mitm server --remote-host 192.168.0.x

To do a full mitm attack, the client should use agent forwarding.

$ ssh -A -p 10022 user@proxyserver

Redirecting client to a honeypot

If the client does not forward an agent, but publickey authentication would be possible on the remote server, SSH-MITM can redirect the session to a honeypot.

$ ssh-mitm server --remote-host 192.168.0.x \
    --enable-auth-fallback \
    --fallback-host HONEYPOT \
    --fallback-username HONEYPOT_USER \
    --fallback-password HONEYPOT_PASSWORD

Connections are only redirected to the honeypot if no agent was forwarded after publickey authentication. All other connections are forwarded to the destination server and a full man in the middle attack is possible.

password authentication - RFC-4252/8

Password authentication is one of the most common login methods. Almost all current operating systems support this method both for local logins and over the network. By default, OpenSSH and many other SSH servers have this type of authentication active.

With SSH, within the encrypted channel, the password is transmitted in clear text. If a client connects to a Man in the Middle server, the server is able to read the username and password in clear text. This information can then be used to log in to other servers if the user exists and uses the same password.

Another problem is that accounts with weak passwords can be compromised relatively easily through a brute force attack. This happens very often with IoT devices because they often have the same username on many devices and they are protected by a default password or only a weak password is set.

On the client side, password authentication should not be used because of security concernes.

Support in SSH-MITM

password authentication is fully supported.

Example SSH-MITM session intercepting password authentication:

$ ssh-mitm server --remote-host 192.168.0.x
2021-09-02 09:51:35,354 [INFO]  starting SSH-MITM 0.5.13
2021-09-02 09:51:38,590 [INFO]  connected client version: SSH-2.0-OpenSSH_8.2p1 Ubuntu-4ubuntu0.3
2021-09-02 09:51:48,629 [INFO]  Client connection established with parameters:
    Remote Address: 192.168.0.x
    Port: 22
    Username: testuser
    Password: secret
    Key: None
    Agent: no agent

keyboard-interactive authentication - RFC-4256

keyboard-interactive is similar to password authentication. The main difference is that the server can send any number of requests to the client, which are necessary for the login process. The server defines both the prompt text and whether the value is visible or not when entered.

In many cases keyboard-interactive is used for 2 factor authentication. In the first step a password is requested and then e.g. the input of a time-based token is necessary (TOTP).

Unless special tools are used to create cryptographically secured input, all input via keyboard-interactive can be reused during a man in the middle attack to login to another server.

Support in SSH-MITM

The current version of SSH-MITM does not support man in the middle attacks using keyboard-interactive authentication.

At the moment only one prompt is sent to the client and the answer is used for password authentication on the remote server.

It’s planned, that the upcoming release of SSH-MITM 1.0, has full support for keyboard-interactive authentication.