Tracking Android/Joker payloads with Medusa, static analysis (and patience)


I recently discovered Medusa and like it very much… for dynamic analysis (I still prefer static analysis, everybody knows that by now?). Medusa is easy to use and comes with a collection of ready-to-use Frida hooks. Launch an Android emulator, a Frida server, install the sample, then launch Medusa python3

use http_communications/uri_logger
use encryption/cipher_1
use code_loading/dump_dyndex
use code_loading/load_class
I use URI hooks (http_communications/uri_logger) in Medusa and see the malware calls those URLs. Android/Joker is known to use URLs such as xxx[.]
Bingo! The look4.oss-ap[..] URL is encrypted. The decryption hooks, encryption/cipher_1, with shows the decrypted value.
DexClassLoader called: /data/user/0/com.designemoji.keyboard/files/audience_network.dex
[+] Dumped /data/user/0/com.designemoji.keyboard/files/audience_network.dex to dump_1
loadClass: com.designemoji.keyboard.EnableActivity
PathClassLoader(f,p) called: /data/user/0/com.designemoji.keyboard/cache/nuff
[+] Dumped /data/user/0/com.designemoji.keyboard/cache/nuff to dump_2
loadClass: seek
DexClassLoader called: /data/user/0/com.designemoji.keyboard/files/seek
[+] Dumped /data/user/0/com.designemoji.keyboard/files/seek to dump_3
DexClassLoader called: /data/user/0/com.designemoji.keyboard/files/Yang
[+] Dumped /data/user/0/com.designemoji.keyboard/files/Yang to dump_4
loadClass: com.xjuys

Loading nuff (payload 1)

DroidLysis doesn’t detect any use of DexClassloader, PathClassLoader or InMemoryDexClassLoader. So, how is the first payload loaded? Let’s locate the URL (look4[…] It is encrypted, so I search where encrypted is used in DroidLysis’ detailed report.

## Cipher
- file=./emojikeyboard.apk-afeb6efad25ed7bf1bc183c19ab5b59ccf799d46e620a5d1257d32669bedff6f/smali/f/a/a/a.smali no= 25 line=b'.method private b()Ljavax/crypto/Cipher;\n'
- file=./emojikeyboard.apk-afeb6efad25ed7bf1bc183c19ab5b59ccf799d46e620a5d1257d32669bedff6f/smali/f/a/a/a.smali no= 63 line=b' invoke-static {v0, v1}, Ljavax/crypto/Cipher;->getInstance(Ljava/lang/String;Ljava/lang/String;)Ljavax/crypto/Cipher;\n'
Code loading the JAR with getClassLoader, then invokes a method named melody()

Static analysis of nuff (payload 1)

The JAR contains a classes.dex with a single class named seek, and a method named melody. It is simple to understand:

  1. It downloads DEX file from https://look4.oss-ap-southeast-5[.]aliyuncs[.]com/nunber
Code of payload 1. Download URL for payload 2 — we also see that class cantus, method bustle is called.

Static analysis of payload 2

Just guess what cantus.bustle() does? It downloads yet another DEX from https://xjuys.oss-accelerate[.] !

Payload 2 is loading … Payload 3

Static analysis of payload 3

This com.xjuys JAR had been already used in several other samples of Joker (sha256: 2edaf2a2d8fd09a254ea41afa4d32b145dcec1ab431a127b2462b5ea58e2903d).

  1. https://xjuys.oss-accelerate[.] We have already seen this payload. It is the same as in this article and contains facebook hooks.
  2. https://beside.oss-eu-west-1[.] It stores the file in the app’s file directory, with filename KBNViao. Then, it loads com.appsflyer.AppsFlyerLib and methods init() then startTracking() [love the name of the method, don’t we? 😏]. This is Apps Flyer SDK, a mobile analytics library.
Connect to remote URL and download payload 4.


The initial DEX is quite heavily obfuscated

  • Payload 1 (designmoji/nuff) has no other use than loading Payload 2
  • Payload 2 (nunber/seek) enables notification listeners (we haven’t detailed this in this article) and loads Payload 3
  • Payload 3 (xjuys/Yang) has yet more malicious code (not detailed here) and loads 2 additional DEX: one for Facebook, the other one contains Apps Flyer SDK.
  • Payload 4a and 4b: Facebook hooks + Apps Flyer SDK.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store



Mobile and IoT malware researcher. The postings on this account are solely my own opinion and do not represent my employer.