A Chinese state-aligned espionage group has become the first documented threat actor to weaponize a known exploit in VS Code in a malicious attack.
Visual Studio Code, or VS Code, is Microsoft’s free source code editor for Windows, Linux, and macOS. According to Stack Overflow’s 2023 survey of 86,544 developers, it’s the most popular integrated development environment (IDE) among both new (78%) and professional developers (74%), by some distance. The next most popular IDE, Visual Studio, was used by 28% of respondents.
In September 2023, a threat researcher described how an attacker could take advantage of a VS Code feature called “Tunnel” to gain initial access to a target’s environment. Initially, the tactic was just fodder for red teaming. Now, according to Palo Alto Networks’ Unit 42, China’s Mustang Panda (aka Stately Taurus, Bronze President, RedDelta, Luminous Moth, Earth Preta, and Camaro Dragon) has used it in an espionage attack against a government entity in southeast Asia.
Dark Reading reached out to Microsoft for comment on this story, with no immediate reply.
Turning VS Code Into a Reverse Shell
“One of the worst fears as a cybersecurity expert is detecting and preventing a signed reverse shell binary,” Truvis Thornton wrote, a whole year prior to Unit 42’s latest research. “Guess what? Microsoft gladly gave us one.”
First introduced in July 2023, VS Code Tunnel allows users to share their VS Code environments on the open Web, and only requires authentication through a GitHub account.
An attacker with their victim’s GitHub credentials could do damage, but much worse is the fact that one can remotely install a portable version of VS Code on a targeted machine. Because it’s a legitimate signed binary, it will not be flagged as suspicious by security software.
And yet, it will walk and talk like a reverse shell. By running the command “code.exe tunnel,” the attacker opens a GitHub authentication page, which they can log into with their own account. Then they’re redirected to a VS Code environment connected to their target’s system, and free to execute commands and scripts and introduce new files at will.
Mustang Panda — a 12-year-old advanced persistent threat (APT) known for espionage against governments, nongovernmental organizations (NGOs), and religious groups in Asia and Europe — used this playbook to perform reconnaissance against its target, drop malware, and, most importantly for its purposes, exfiltrate sensitive data.
How to Deal with VSCode
“While the abuse of VSCode is concerning, in our opinion, it is not a vulnerability,” Assaf Dahan, director of threat research for Unit 42, clarifies. Instead, he says, “It’s a legitimate feature that was abused by threat actors, as often happens with many legitimate software (take lolbins, for example).”
And there are a number of ways organizations can protect against a bring-your-own-VSCode attack. Besides hunting for indicators of compromise (IoCs), he says, “It’s also important to consider whether the organization would want to limit or block the use of VSCode on endpoints of employees that are not developers or do not require the use of this specific app. That can reduce the attack surface.”Â
“Lastly, consider limiting access to the VSCode tunnel domains ‘.tunnels.api.visualstudio[.]com’ or ‘.devtunnels[.]ms’ to users with a valid business requirement. Notice that these domains are legitimate and are not malicious, but limiting access to them will prevent the feature from working properly and consequently make it less attractive for threat actors,” he adds.
A Second, Overlapping Attack
While investigating the Mustang Panda attack, Unit 42 came across a second threat cluster occupying the same target’s systems.
In this case, the attacker abused imecmnt.exe — a legitimate and signed file associated with Microsoft’s Input Method Editor (IME), used for generating text in languages not conducive to the QWERTY keyboard — with some dynamic link library (DLL) sideloading. The file they dropped, ShadowPad, is a 7-year-old modular backdoor popular among Chinese threat actors.
This compromise occurred at the same time as the VS Code exploitation, often on the same endpoints, and the overlaps didn’t end there. Still, researchers couldn’t say for certain whether this second cluster of malicious activity could be attributed to Mustang Panda. “There could also be other possible scenarios to explain this connection,” they wrote. “For example, it could be a joint effort between two Chinese APT groups or perhaps two different groups piggybacking on each other’s access.”
Don’t miss the latest Dark Reading Confidential podcast, where we talk to two cybersecurity professionals who were arrested in Dallas County, Iowa, and forced to spend the night in jail — just for doing their pen-testing jobs. Listen now!