The DuoConnect client enables users to establish SSH connections to hosts protected by a DNG instance. When a user initiates an SSH connection to a DNG-protected host for the first time using DuoConnect, the user’s browser is opened to a login screen in order to complete authentication determined by the contents of the -relay argument. If the ‘-relay’ is set to a URL beginning with http://, then the browser will initially attempt to load the URL over an insecure HTTP connection, before being immediately redirected to HTTPS (in addition to standard redirect mechanisms, the DNG uses HTTP Strict Transport Security headers to enforce this). After successfully authenticating to a DNG, DuoConnect stores an authentication token in a local system cache, so users do not have to complete this browser-based authentication workflow for every subsequent SSH connection. These tokens are valid for a configurable period of time, which defaults to 8 hours. If a user running DuoConnect already has a valid token, then instead of opening a web browser, DuoConnect directly contacts the DNG, again using the configured -relay value, and sends this token, as well as the intended SSH server hostname and port numbers. If the -relay argument begins with http://, then this request will be sent over an insecure connection, and could be exposed to an attacker who is sniffing the traffic on the same network. The DNG authentication tokens that may be exposed during SSH relay may be used to gain network-level access to the servers and ports protected by that given relay host. The DNG provides network-level access only to the protected SSH servers. It does not interact with the independent SSH authentication and encryption. An attacker cannot use a stolen token on its own to authenticate against a DNG-protected SSH server.
The product transmits sensitive or security-critical data in cleartext in a communication channel that can be sniffed by unauthorized actors.
| Name | Vendor | Start Version | End Version |
|---|---|---|---|
| Duoconnect | Duo | * | 1.1.1 (excluding) |
Many communication channels can be “sniffed” (monitored) by adversaries during data transmission. For example, in networking, packets can traverse many intermediary nodes from the source to the destination, whether across the internet, an internal network, the cloud, etc. Some actors might have privileged access to a network interface or any link along the channel, such as a router, but they might not be authorized to collect the underlying data. As a result, network traffic could be sniffed by adversaries, spilling security-critical data. Applicable communication channels are not limited to software products. Applicable channels include hardware-specific technologies such as internal hardware networks and external debug channels, supporting remote JTAG debugging. When mitigations are not applied to combat adversaries within the product’s threat model, this weakness significantly lowers the difficulty of exploitation by such adversaries. When full communications are recorded or logged, such as with a packet dump, an adversary could attempt to obtain the dump long after the transmission has occurred and try to “sniff” the cleartext from the recorded communications in the dump itself.