IPV Spyware Discovery (ISDi) Tool

Checks Android or iOS devices for apps used to surveil or track victims ("stalkerware", "spouseware", "spyware"). ISDi's technical details are included in "Clinical Computer Security for Victims of Intimate Partner Violence" (USENIX 2019). The blacklist is based on apps crawled in "The Spyware Used in Intimate Partner Violence" (IEEE S&P 2018).

Also checks for signs of jailbroken iOS devices or rooted Android devices.

Installing ISDi :computer:

Right now, ISDi currently only natively supports macOS and Linux. If you are using a Windows device, you can use the Windows Subsystem for Linux 2 (WSL2), which can be installed by following these instructions. After this, follow the remaining instructions as a Linux user would, cloning/running ISDi inside the Linux container of your choice.

Python dependencies

You will need Python 3.6 or higher (check by running python3 in your Terminal and see what happens). On macOS, you can get this by running the following commands in your Terminal application: xcode-select --install (installs developer tools); followed by /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)" to get Brew (a software package manager); finally, brew install python to get Python 3.6+.
Run pip3 install -r requirements.txt in the base directory of this repository to get the required Python modules.

Miscellaneous System dependencies

Install ADB for your operating system (via https://developer.android.com/studio/releases/platform-tools.html). On macOS, try brew cask install android-platform-tools. On Debian-based systems, try sudo apt install adb. Windows via WSL2 only: Installing adb is not so straightforward in WSL2, and it won't work straightaway. You have to ensure having the same version of adb both in WSL2 and in normal Windows (with adb version), then you will need to start the adb process first in Windows, then in WSL2 (with for example adb devices).
Install expect. On macOS, run brew install expect. On Debian-based systems, run sudo apt install expect.
Linux/WSL2 only: If you are running Linux or WSL2, install patchelf (on Debian-based systems, this can be done like sudo apt install patchelf) and then run ./scripts/patch_ios_dylibs.sh. You may need to re-run this script if you are collaborating and other ISDi users overwrite the iOS binaries with their own absolute shared object paths. More details here (including for macOS users).

Blacklist dependencies

Fill out https://forms.gle/qzres5fQaRbwsyni7 with a legitimate request for the blacklist ISDi requires to run. Please note that at this time, we are only considering requests from those working for victim services organizations. Approved requests will receive a follow-up email to the address entered on the form, so be sure to check your inbox. We will provide static_data.zip, which should be unzipped and replacing the public facing static_data directory in this repository.

Running ISDi

After ISDi is installed, with an Android or iOS device plugged in and unlocked, run the following command in the terminal (in the base directory of this repository)

$ bash run.sh

Then navigate to http://localhost:5000 in the browser of your choice (or http://localhost:5002 if in test mode). You will see ISDi running as a web app. Click on "Scan Instructions" and follow the instructions to prepare your device for the scan.

It should look something like this:

Phone Scanner UI before scan

Connect a device and click on the suitable button Android or iOS. Give it a nickname and click "Scan now". (Please connect one device at a time.) It will take a few seconds for the scan to complete. We are working to have all scan results done at once on Android, but for the time being please leave the device plugged in when clicking on apps on the scan results table.

After the scan, the UI will look something like this:

Phone Scanner UI after scan

Consultation form data

ISDi is intended to be used by advocates for victims of intimate partner violence in a clinical setting; you can add detailed notes about a victim's tech abuse situation by clicking "Start Consult Form" on ISDi's homepage. The results will be saved in data/fieldstudy.db and can be viewed/edited by navigating to /form/edit.

Some consult form data may not be relevant for use in other organizations (e.g., the meeting location being in a borough of New York City). Please consider adapting the form for your needs. One can do this by modifying the Client class in server.py and use sa.create_all() (sa is obtained by wrapping SQLAlchemy over the Flask app) to obtain the new schema. Then place the new schema in schema.sql by updating the clients_notes table.

Debugging tips

If you encounter errors, please file a GitHub issue with the server error output (or send it as an email to havron@cs.cornell.edu). Pull requests are welcome. Inspect apps on the device manually if you cannot resolve a failure.

Android tips

In the terminal of the computer, run adb devices to see if the device is connected properly.

iOS tips

In the terminal of the computer (in the base directory of this repository), run ./static_data/libimobiledevice-darwin/idevice_id -l to see if the device is connected properly (replace darwin with linux if your system is Linux.)

Cast iOS Screens or Mirror Android Screens

It is possible to view your device screen(s) in real time on the macOS computer in a new window. This may be useful to have while you are running the scan (and especially if you use the privacy checkup feature), as it will be easy for you to see the mobile device screen(s) in real time on the Mac side-by-side with the scanner.

How to do it: you can mirror Android device screens in a new window using scrcpy, and cast iOS device screens on macOS with QuickTime 10 (launch it and click File --> New Movie Recording --> (on dropdown by red button) the iPhone/iPad name).

Downloaded data

The data downloaded and stored in the study are the following. 1. A sqlite database containing the feedback and actions taken by the user. 2. phone_dump/ folder will have dump of some services in the phone. (For Android I have figured out what are these, for iOS I don't know how to get those information.)

Android

The services that we can dump safely using dumpsys are the following.

Application static details: package Sensor and configuration info:
location, media.camera, netpolicy, mount Resource information:
cpuinfo, dbinfo, meminfo Resource consumption: procstats,
batterystats, netstats, usagestats App running information: activity,
appops

See details about the services in notes.md

iOS

Only the appIds, and their names. Also, I got "permissions" granted to the application. I don't know how to get install date, resource usage, etc. (Any help will be greatly welcomed.)

Code structure

phone_scanner.py has all the logic required to communicate with Android and iOS devices.
parse_dump.py has all of the logic required to extract dumped info from the devices. After the initial scan, the server will rely on this parser rather than needing an active connection to the device (work in progress). For now, please keep your device plugged in when looking at scan results.
server.py is the Flask web server.
templates/ folder contains the html templates rendering in the UI
webstatic/ folder contains the .css and .js files
phone_dumps/ folder will contain the data recorded from the phone (as well as in data/fieldstudy.db.

TODO 1.

https://docs.google.com/document/d/1fy6RTo9Gc0rBUBHAhKfSmqI99PSPCBsAdEUIbpGIkzQ/edit

~How to figure out off-store apps in Android and iOS? Check the installer in adb shell pm packages -i~ 3. For iOS, how to find out app installation dates, resource usage, etc? 4. Explore viability of WebUSB and WebADB.

See notes.md for other developer helps.