sebastianevan200-stack/veglia — explained in plain English
Analysis updated 2026-05-18
Let your AI assistant check what is currently shown on your own phone screen.
Set up a consent-based screen-sharing bridge between two partners.
Run a self-hosted alternative to cloud screen-mirroring services.
Trigger a custom notification script whenever a new screenshot arrives.
| sebastianevan200-stack/veglia | justson/pixel-activity-keep-alive | setsu2420/personal-knowledgebase | |
|---|---|---|---|
| Stars | 27 | 27 | 27 |
| Language | Java | Java | Java |
| Last pushed | — | 2018-12-22 | — |
| Maintenance | — | Dormant | — |
| Setup difficulty | moderate | easy | hard |
| Complexity | 3/5 | 2/5 | 4/5 |
| Audience | vibe coder | developer | researcher |
Figures from each repo's GitHub metadata at analysis time.
Requires a reachable server you control, a generated secret token, and manually enabling an Android accessibility permission.
Veglia is a small self-hosted tool that lets an AI assistant look at what is currently on your phone's screen, with your explicit permission. The name comes from an Italian word meaning to keep watch through the night over someone you love, and the project frames itself around consent: it is meant for two people who both agreed to it, not for secretly watching someone's phone. The README states plainly that using it without the phone owner's knowledge is abuse of the person and of the project. The system has three parts: an Android app installed on the phone being watched, a small Python server you run yourself on a machine you control, and a one-file command line tool your AI can run. When the AI asks what is on the phone right now, it runs a command that tells the app to capture the screen and upload it to your server. The server can then fire a notifier of your choosing to deliver that image into your AI's conversation. According to the README, several design choices support consent and privacy: no cloud service is involved, all data stays on your own server. Only the most recent screenshots are kept, five by default, with older ones deleted as new ones arrive. The phone owner always sees a visible notification while the app runs, must manually turn on an accessibility permission for it to work, and can stop it at any time. Every request to the server requires a secret token you generate yourself. This only works on Android, using a screen capture feature built into Android's accessibility tools starting with Android 11. Apple's iOS does not allow this kind of background screen capture, so there is no iPhone version. Setup involves running the Python server on a reachable machine, building and installing the Android app with your server address and token, and connecting a script that hands the screenshot to your AI. The project is released under the MIT license.
A consent-based, self-hosted bridge that lets an AI assistant view a phone's screen when both people have agreed to it.
Mainly Java. The stack also includes Android, Python, Java.
MIT license: free to use, modify, and distribute, including commercially, as long as the copyright notice is kept.
Setup difficulty is rated moderate, with roughly 1h+ to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Verify against the repo before relying on details.