Analysis updated 2026-05-18
Explore how to combine screen automation, computer vision, and an AI judge into a working phone-app control loop, as an engineering exercise.
Automatically screenshot and evaluate dating profiles against a rubric you write, then like or skip them.
Generate a personalized opening message in a configured voice when a profile is liked.
Analyze your own dating profile with scan_self.py and get a Markdown report with photo and prompt suggestions.
| terrabyte-dev/hinge-auto | aaravkashyap12/advise-project-approach | abu-rayhan-alif/django-saas-kit | |
|---|---|---|---|
| Stars | 23 | 23 | 23 |
| Language | Python | Python | Python |
| Setup difficulty | hard | easy | moderate |
| Complexity | 4/5 | 2/5 | 3/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires Android Studio, a configured emulator, ADB, and either a paid Claude API key or a locally running Ollama model, violates Hinge's Terms of Service.
HingeAuto is an experimental tool that automates the Hinge dating app by running it inside an Android emulator on your computer and using an AI model to make decisions about each profile. The README opens with a clear warning: this violates Hinge's Terms of Service and carries a real risk of a permanent account ban with no appeal. The project is described as an educational exercise in connecting computer vision and AI judgment to a phone app interface, not as practical dating advice. The setup involves installing Android Studio, creating a virtual Android device, and installing Hinge on it the same way you would on a real phone. Once the emulator is running, the project uses ADB (the standard Android debugging tool) to take screenshots and send tap and scroll commands to the emulator's screen. For each profile the bot encounters, it scrolls from top to bottom, captures the screen, and sends the images to an AI model along with a rubric you write yourself. The rubric is a plain-text file called a mode that describes what should and should not receive a like. Based on the AI's judgment, the bot either taps Skip or taps the heart and asks the AI to write a personalized opening message in a voice you configure. Two AI options are supported: the Claude API (paid, higher quality) or a locally running Ollama model (free, lower quality). The README estimates roughly $0.25 per day using Claude at the free-tier like cap of 8 profiles. A separate script called scan_self.py uses the same screenshot-and-AI approach on your own profile, the view other users actually see, and writes a Markdown report with specific suggestions for photos and prompts. The README notes this is the only feature that does not touch swiping or messaging and is therefore the one part that does not violate Hinge's terms. Calibration scripts handle differences between emulator screen sizes if your setup uses a different device than the default Pixel 10 at 1080 by 2424 resolution.
An experimental Android-emulator bot that uses AI vision to judge Hinge dating profiles and write openers, explicitly against Hinge's Terms of Service.
Mainly Python. The stack also includes Python, ADB, Android Studio.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.