Urgent.News

What's breaking now, across thousands of outlets.

Tech

I built a ghost that haunts my desktop until it can see my work

My submission for the Hacktoberfest Open-Source AI Challenge: Ghost in the Machine. Gemma 3 4B, Ollama and PySide6. Nothing leaves my laptop. A reminder asks one thing of you: a tap. Snooze, dismiss, "done". None of them checks anything and I am very good at tapping. So I built Haunt, a small ghost that appears over whatever I am doing, says something unsettling and does not leave until it can…

A ghost named Haunt now haunts the user's desktop, appearing over their tasks and speaking unsettling lines until the work is completed. Created for the Hacktoberfest Open-Source AI Challenge, Haunt uses Gemma 3 4B, Ollama, and PySide6 to evaluate the user's work through screenshots. The ghost appears after a tap, asking if the work is done, and becomes increasingly agitated if ignored.

If the screenshot does not show the completed task, Haunt provides evidence and asks the user to prove the work. The model checks the evidence, and if it is insufficient, the ghost remains and prompts the user to try again. Haunt only captures visual evidence if necessary, as it is more cost-effective than the model interpreting pixels.

The ghost's buttons did not respond to clicks due to the Qt.Tool window flag, which was removed to fix the issue. The Ollama host was also misconfigured initially, causing connectivity problems. The ghost uses the Windows speech engine for its voice, running locally on the user's machine without accessing external servers or APIs.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

Rutba Vision: video surveillance for camera networks past the size one recorder copes with

First published on rutba.io . A handful of cameras is a product you buy. A hundred and fifty, writing terabytes a day, is an operations problem, and it fails the same ways each time.

  • Rutba Vision handles large camera networks without single recorder limitations
  • Software runs node agents on all machines for edge, analysis, and core roles
  • Footage recorded in self-contained fMP4 segments with SHA-256 hashing

More from Saturday 31 October →