Reverse engineering a Smarter Coffee machine for fun and a security lesson

Pierluigi Paganini October 10, 2016

Simone Margaritelli has done a reverse engineering of the Smarter Coffee IoT Machine Protocol to control the machine from his terminal. What is the lesson?

While security industry is stressing the need to adopt a security by design approach for IoT devices, security researchers continue to find flawed and poorly designed smart objects.

Clearly, such kind of devices is a privileged target for crooks that could abuse them to conduct a wide range of illegal activities.

Ok … but it’s time for a coffee break now and surfing the web I found a curious and interesting article of a popular Italian hacker, Simone Margaritelli, aka evilsocket.  Simone is a former blackhat hacker now mobile security researcher and senior ASM/C/C++ developer for Zimperium firm, he is the creators of the popular tool bettercap.

Like me, Simone loves coffee so a few days ago he bought a Smarter Coffee machine that can be controlled via a mobile application that allows users to prepare a good coffee with many options.

Simone Margaritelli decided to do a reverse engineering of the Smarter Coffee IoT Machine Protocol with the intent of control the coffee machine even from his terminal.


The expert focused its analysis on classes and methods present in the source code of the app, then he found something of interest in the am.smarter.smarterandroid.models.a class.

Smarter Coffee iot-machine-protocol

The researcher discovered the way the app and the machine communicate and which is the protocol they use.

“Each of these “packets” is sent to tcp port 2081 of the machine, the protocol is very simple.

First byte: the command number.
Second byte to N: optional data ( depending on the command code ).
Last byte: always 0x7e which indicates the end of the packet.
Responses can vary, but for most of the commands they are:

First byte: response size
Second byte: status ( 0 = success otherwise error code )
Last byte: always 0x7e.
An example command and response, the one to keep the coffee warm for 5 minutes for instance, would be:

COMMAND : 0x3e 0x05 0x7e
RESPONSE : 0x03 0x00 0x7e”

At this point, it was a joke for Simone to write a simple console to send commands to the Smarter Coffee machine as you can see in the video PoC published by the hacker.

Simone has published the code on GitHub, below a few sample of the commands available to control the coffee machine.

  • Make one cup of coffee: coffee make.
  • Make two cups using the filter instead of the beans in the grinder coffee make –filter.
  • Keep coffee warm for ten minutes coffee warm –keep-warm=10.

Simone Margaritelli highlighted that anyone on the same network of the machine could send commands to the device due to the absence of authentication.

“Even if the mobile app requires you to register an account, access to port 2081 is completely unauthenticated ( in fact, I’ve found that the user account is only used for statistics using the Firebase API ), anyone on your network could access it and even flash a new firmware with no authentication required ( I reversed the UPDATE_FIRMWARE packet as well but you won’t find it on the repo 😛 )” Simone wrote in a blog post.

[adrotate banner=”9″]

Pierluigi Paganini

(Security Affairs – Smarter Coffee, IoT)

you might also like

leave a comment