<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Shrishti Soni</title>
    <description>The latest articles on DEV Community by Shrishti Soni (@still_storm).</description>
    <link>https://dev.to/still_storm</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4158398%2F43bfd494-189b-4c22-a035-7f5d84d8d1da.jpg</url>
      <title>DEV Community: Shrishti Soni</title>
      <link>https://dev.to/still_storm</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/still_storm"/>
    <language>en</language>
    <item>
      <title>GhOStly</title>
      <dc:creator>Shrishti Soni</dc:creator>
      <pubDate>Thu, 29 Oct 2026 14:25:50 +0000</pubDate>
      <link>https://dev.to/still_storm/ghostly-52ki</link>
      <guid>https://dev.to/still_storm/ghostly-52ki</guid>
      <description>&lt;p&gt;*What I Built&lt;br&gt;
GhOStly (Ghost + OS) is a horror mini-game where your own desktop becomes the haunted level.&lt;br&gt;
A tiny ghost floats on top of your screen with one button: START. Before pressing it, you can drag the ghost onto any app, window or folder you like. The moment you press START, the game takes a screenshot of your screen, finds every icon and every piece of text, and turns each one into a monster with glowing eyes and teeth. Then a little exorcist hero appears in front of you. You control him with the keyboard, aim a magic lantern with the mouse, and purify every monster. A purified monster turns back into a normal icon.&lt;/p&gt;

&lt;p&gt;How I interpreted the theme: "Ghost in the Machine", taken literally. The ghost lives inside your own computer (inside the OS), and the haunted house is the screen you use every day. The twist at the end goes one step further: the real ghost might not be in the machine at all.&lt;/p&gt;

&lt;p&gt;Code&lt;br&gt;
&lt;/p&gt;
&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/still4shrishti" rel="noopener noreferrer"&gt;
        still4shrishti
      &lt;/a&gt; / &lt;a href="https://github.com/still4shrishti/ghOStly" rel="noopener noreferrer"&gt;
        ghOStly
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;👻 ghOStly&lt;/h1&gt;
&lt;/div&gt;
&lt;p&gt;An eerie and innovative &lt;strong&gt;2D Action-Shooter Game&lt;/strong&gt; where your own desktop environment turns against you! Upon launch, the game captures a snapshot of your screen and mutates your apps, folders, and text into hostile anomalies.&lt;/p&gt;
&lt;p&gt;Will you exorcise the ghosts haunting your desktop, or will you succumb to the monster within?&lt;/p&gt;

&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;🔥 Features&lt;/h2&gt;
&lt;/div&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Desktop Camouflage Mechanic:&lt;/strong&gt; The game dynamically crops your real desktop icons and turns them into floating, twitching entities.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The Little Exorcist (Hero):&lt;/strong&gt; Control a robed wizard holding a glowing lantern that illuminates the dark voids left by your missing desktop items.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Intense Final Boss Fight:&lt;/strong&gt; Once the minor entities are cleared, the desktop fragments converge into the center to assemble &lt;strong&gt;"The Desktop Beast"&lt;/strong&gt;—a towering humanoid titan plated with your own icons.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Cinematic Horror Twist:&lt;/strong&gt; A scripted atmospheric ending sequence that shifts the focus directly onto the player.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Immersive Visuals:&lt;/strong&gt; Features dynamic lighting glows, active…&lt;/li&gt;
&lt;/ul&gt;&lt;/div&gt;
  &lt;/div&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/still4shrishti/ghOStly" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;/div&gt;


&lt;p&gt;How I Built It&lt;br&gt;
Python + pygame for the game loop, graphics, hero animation, hearts and sound&lt;br&gt;
mss to capture the screenshot&lt;br&gt;
OpenCV + NumPy to find icon and text shaped blobs (edge detection and dilation), which are then cut out and turned into monsters&lt;br&gt;
PyInstaller to package the game as a Windows .exe&lt;br&gt;
I used an AI coding assistant (Claude) to help write and debug parts of the code.&lt;br&gt;
Open-source AI part:&lt;/p&gt;

&lt;p&gt;Why Does Open Innovation Matter?&lt;br&gt;
This game only exists because of open-source tools. pygame, OpenCV, mss and NumPy are free, inspectable and well documented, so a small team could build a full game in a hackathon.&lt;/p&gt;

&lt;p&gt;It also matters for privacy. This game reads your real screen, which can contain private things. Because everything runs locally on your own machine, the screenshot stays in memory and is never uploaded anywhere. A closed cloud API would have meant sending pictures of your desktop to someone else's server.&lt;br&gt;
The Spirit of It&lt;br&gt;
The part I care about most is the ending. After you purify the last monster, the game asks: "How will you kill the monster inside you?" The screen turns dark, the hero walks to the middle, looks straight at you, his eyes turn red, and the screen goes black. Then the program closes and your desktop returns to normal, as if nothing happened.&lt;br&gt;
I wanted the scariest place in the game to be somewhere you already trust: your own screen.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>SINK</title>
      <dc:creator>Shrishti Soni</dc:creator>
      <pubDate>Sat, 24 Oct 2026 17:36:26 +0000</pubDate>
      <link>https://dev.to/still_storm/sink-2lfn</link>
      <guid>https://dev.to/still_storm/sink-2lfn</guid>
      <description>&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;I built &lt;strong&gt;SINK, a JARVIS-style local AI desktop assistant&lt;/strong&gt; that lets users control their PC using voice commands and interact with an AI assistant without relying entirely on cloud services.&lt;/p&gt;

&lt;p&gt;SINK can open applications, control the mouse and keyboard, scroll, type, manage tasks and reminders, answer questions, assist with studying, and interact with YouTube. It also has an animated robot face that reacts to conversations.&lt;/p&gt;

&lt;p&gt;The goal is simple: make everyday computer interactions faster, more natural, and hands-free.&lt;/p&gt;

&lt;p&gt;Instead of manually switching between applications and repeating routine actions, users can simply say, "SINK, open Chrome" or "SINK, search YouTube for Python tutorials."&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Demo highlights:&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Voice activation using the wake word "SINK".&lt;/li&gt;
&lt;li&gt;Opening applications and controlling the desktop.&lt;/li&gt;
&lt;li&gt;Searching YouTube and playing the first video.&lt;/li&gt;
&lt;li&gt;Asking AI questions and getting study assistance.&lt;/li&gt;
&lt;li&gt;Using reminders, timers, and task management.&lt;/li&gt;
&lt;li&gt;Watching the animated robot respond to commands.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Project:&lt;/strong&gt; SINK, Local Voice AI Desktop Assistant&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Repository:&lt;/strong&gt; &lt;a href="https://github.com/still4shrishti/SINK-" rel="noopener noreferrer"&gt;https://github.com/still4shrishti/SINK-&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Tech stack:&lt;/strong&gt; Python, PySide6, Faster-Whisper, Qwen 3, LM Studio, PyAutoGUI, and pyttsx3.&lt;/p&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;SINK combines open-source tools and locally running AI to create a voice-controlled desktop assistant.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Brain:&lt;/strong&gt; Qwen 3 running locally through LM Studio, providing AI-powered responses.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Ears:&lt;/strong&gt; Faster-Whisper converts spoken commands into text.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Voice:&lt;/strong&gt; pyttsx3 converts responses into speech.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Interface:&lt;/strong&gt; PySide6 powers the desktop window and animated robot face.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hands:&lt;/strong&gt; PyAutoGUI handles mouse movements, keyboard input, clicking, and scrolling.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;System monitoring:&lt;/strong&gt; psutil provides information such as CPU usage, RAM, battery, and disk status.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;I built the assistant around practical commands, allowing users to interact with their computer through natural voice instructions instead of relying only on traditional interfaces.&lt;/p&gt;

&lt;p&gt;The project also includes productivity features such as tasks, reminders, timers, clipboard assistance, dictation, and study quizzes.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;Open innovation makes projects like SINK possible by allowing developers to combine existing open-source tools and open-weight AI models into something useful and personal.&lt;/p&gt;

&lt;p&gt;Running Qwen 3 locally through LM Studio gives SINK an AI brain without requiring every conversation to be sent to a paid cloud AI API. Faster-Whisper enables local speech recognition, while Python libraries provide the tools needed for desktop automation.&lt;/p&gt;

&lt;p&gt;This approach offers more control over privacy, customization, and operating costs. It also lets developers experiment, modify components, and build new capabilities without depending entirely on a single closed platform.&lt;/p&gt;

&lt;p&gt;For me, open innovation means taking powerful technologies that already exist and combining them into something people can actually use.&lt;/p&gt;

&lt;h2&gt;
  
  
  My Agent Session
&lt;/h2&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/TtrTHy99Ceg" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;h2&gt;
  
  
  The Lazy Payoff
&lt;/h2&gt;

&lt;p&gt;SINK is designed to reduce the small, repetitive tasks that interrupt work and study.&lt;/p&gt;

&lt;p&gt;Instead of manually opening a browser, searching for a tutorial, switching windows, or navigating through multiple controls, users can issue a voice command and let SINK handle supported actions.&lt;/p&gt;

&lt;p&gt;For example:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;"SINK, open Chrome."&lt;/li&gt;
&lt;li&gt;"SINK, search YouTube for Python tutorials."&lt;/li&gt;
&lt;li&gt;"SINK, remind me in 10 minutes to drink water."&lt;/li&gt;
&lt;li&gt;"SINK, explain what I copied."&lt;/li&gt;
&lt;li&gt;"SINK, quiz me on binary trees."&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The real payoff is making everyday computer use feel more conversational and reducing the effort needed to complete routine tasks.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>SINK</title>
      <dc:creator>Shrishti Soni</dc:creator>
      <pubDate>Sat, 03 Oct 2026 19:00:00 +0000</pubDate>
      <link>https://dev.to/still_storm/sink-5ha3</link>
      <guid>https://dev.to/still_storm/sink-5ha3</guid>
      <description>&lt;p&gt;I built &lt;strong&gt;SINK&lt;/strong&gt;, a voice assistant for Windows that lets you control your laptop without using your hands.&lt;/p&gt;

&lt;p&gt;I built it for my friend mahima. she can't use her hands properly because she applies henna (mehdi in indian language), so basic things like opening a browser, clicking a button, scrolling or searching a homework answer became very hard. Windows has voice control, but it only follows commands. It can't see what's on the screen or help with the actual work.&lt;/p&gt;

&lt;p&gt;With SINK, you say "SINK" and then a command:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Control the laptop:&lt;/strong&gt; "open Chrome", "scroll down", "type hello", "search Python tutorials". It starts listening on launch, so no mouse is needed. It understands some Hinglish commands too.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Click by name:&lt;/strong&gt; "SINK click Send" finds the button through Windows UI Automation and clicks it, with no pixel guessing.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;See the screen:&lt;/strong&gt; "SINK what's on my screen" and "SINK explain this error" use a local vision model.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Solve copied questions:&lt;/strong&gt; "SINK solve what I copied" opens a page with properly rendered maths and a &lt;strong&gt;Final Answers&lt;/strong&gt; box at the top. Or say "in Notepad" for plain text. It always warns: &lt;em&gt;"Please check it once, I can make mistakes."&lt;/em&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Show it's working:&lt;/strong&gt; a small chibi robot shows "SOLVING..." with moving dots, so [friend's name] knows it hasn't frozen.&lt;/li&gt;
&lt;li&gt;Dictation, reminders, to-do notes and a small memory, all by voice.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  screenshot of icon
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzxg4hgnd121di1l5biyr.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzxg4hgnd121di1l5biyr.png" alt=" " width="210" height="272"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Flssz6qk5ghrr9j9aocs4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Flssz6qk5ghrr9j9aocs4.png" alt=" " width="232" height="288"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;


&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/still4shrishti" rel="noopener noreferrer"&gt;
        still4shrishti
      &lt;/a&gt; / &lt;a href="https://github.com/still4shrishti/SINK-" rel="noopener noreferrer"&gt;
        SINK-
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      SINK | A Python-based AI desktop assistant that listens, talks, automates tasks, and reacts through an interactive virtual character.
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;SINK: Local Voice AI Desktop Assistant&lt;/h1&gt;
&lt;/div&gt;
&lt;p&gt;SINK is a Jarvis-style voice assistant that runs &lt;strong&gt;fully on your own PC&lt;/strong&gt;
Say "SINK" and give a command, and it will open apps, click, scroll, type, set reminders, play games and chat with you
It has a small animated robot face that reacts to what you say.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Brain:&lt;/strong&gt; Qwen 3, running locally through LM Studio&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Ears:&lt;/strong&gt; faster-whisper (speech to text, on CPU)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Voice:&lt;/strong&gt; pyttsx3 (text to speech)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Face and window:&lt;/strong&gt; PySide6 (Qt)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hands:&lt;/strong&gt; pyautogui (mouse and keyboard control)&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Your voice and notes are not sent to any cloud service. Everything stays on your computer.&lt;/p&gt;

&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Features&lt;/h2&gt;

&lt;/div&gt;
&lt;div class="markdown-heading"&gt;
&lt;h3 class="heading-element"&gt;Talk to it&lt;/h3&gt;

&lt;/div&gt;
&lt;ul&gt;
&lt;li&gt;Wake word: say &lt;strong&gt;"SINK"&lt;/strong&gt; (also hears "sync", "think", "zinc"). Example: &lt;em&gt;"SINK open Chrome"&lt;/em&gt;
&lt;/li&gt;
&lt;li&gt;Hands-free listening toggle, or type commands in the text box&lt;/li&gt;
&lt;li&gt;Answers general questions using Qwen 3 from LM Studio&lt;/li&gt;
&lt;li&gt;Chain commands: &lt;em&gt;"press click and then scroll down"&lt;/em&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;div class="markdown-heading"&gt;
&lt;h3 class="heading-element"&gt;Control the PC&lt;/h3&gt;

&lt;/div&gt;
&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Say&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;/table&gt;&lt;/div&gt;…&lt;/div&gt;
  &lt;/div&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/still4shrishti/SINK-" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;/div&gt;


&lt;p&gt;Setup: pip install reqirements, start LM Studio's local server with a vision model, then run &lt;code&gt;python gui.py&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;Everything runs on the laptop. Nothing is sent to a cloud service.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Speech to text:&lt;/strong&gt; Whisper through &lt;code&gt;faster-whisper&lt;/code&gt;, running locally on CPU.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Language and vision model:&lt;/strong&gt; open-weight models served locally through LM Studio's OpenAI-compatible API ([model names, e.g. Qwen3-VL for screen reading and solving]).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Interface:&lt;/strong&gt; Python with PySide6: the main window, particle "core", and the chibi companion.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Computer control:&lt;/strong&gt; &lt;code&gt;pyautogui&lt;/code&gt; for mouse and keyboard, &lt;code&gt;pywinauto&lt;/code&gt; (Windows UI Automation) to click buttons by name.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Voice output:&lt;/strong&gt; &lt;code&gt;pyttsx3&lt;/code&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Parts I had to solve myself:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Mishearing:&lt;/strong&gt; Whisper often mishears short commands ("click" vs "tap"), so I wrote a layer that maps many ways of saying something to one command.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Maths display:&lt;/strong&gt; small models write maths in LaTeX, which looks broken in Notepad. I added a converter to plain symbols and a browser page that renders the maths properly.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Safety:&lt;/strong&gt; a blocked-terms list for protected system areas, and a mistake warning on every solution.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;My friend's screen can show private things like messages, school work and logins. Because the models are open-weight and run locally, &lt;strong&gt;that screen content never leaves the laptop&lt;/strong&gt;. With a closed cloud API I would have been sending screenshots of a real person's screen to a third party, which I wouldn't be comfortable asking my friend to accept.&lt;/p&gt;

&lt;p&gt;It also means no API bills, no rate limits, and no dependence on a company's pricing or policy. A student with a normal laptop can build an accessibility tool for someone they know. And because the tools (Whisper, local models, PySide6, pywinauto) are open, I could combine them in ways no single product offered.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>weekendchallenge</category>
      <category>hf26challenge</category>
    </item>
  </channel>
</rss>
