<?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: Harsh Lone</title>
    <description>The latest articles on DEV Community by Harsh Lone (@harshlone14).</description>
    <link>https://dev.to/harshlone14</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%2F4179069%2Fa4add28f-50b0-4a53-91c6-c523b5eaa5fb.jpg</url>
      <title>DEV Community: Harsh Lone</title>
      <link>https://dev.to/harshlone14</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/harshlone14"/>
    <language>en</language>
    <item>
      <title>HauntedShell: When Your Local Terminal Starts Whispering Back at Midnight</title>
      <dc:creator>Harsh Lone</dc:creator>
      <pubDate>Wed, 28 Oct 2026 18:46:39 +0000</pubDate>
      <link>https://dev.to/harshlone14/hauntedshell-when-your-local-terminal-starts-whispering-back-at-midnight-48hp</link>
      <guid>https://dev.to/harshlone14/hauntedshell-when-your-local-terminal-starts-whispering-back-at-midnight-48hp</guid>
      <description>&lt;p&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week4-2026-10-26"&gt;Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine&lt;/a&gt;*&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;For this week's Ghost in the Machine challenge, I wanted to build something that bridges the nostalgic , feeling of late-night coding sessions with practical local AI. Meet HauntedShell—a lightweight Python tool that "possesses" your local shell session.&lt;/p&gt;

&lt;p&gt;Whenever you type commands or prompts into the terminal, an open-weight language model running locally intercepts your input and responds with strange, atmospheric, and oddly philosophical commentary. Complete with spooky ASCII art, custom colored text, and a slow typewriter effect, it turns an ordinary CLI window into an interactive digital ghost story.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&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%2Fhm29w6yidq4n388jzqk6.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%2Fhm29w6yidq4n388jzqk6.PNG" alt="HAUNTEDSHELL CLI TERMINAL DEMO AND INTERACTIVE GHOST OUTPUT" width="800" height="520"&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%2Fepmjag8e4m37z01slgct.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%2Fepmjag8e4m37z01slgct.png" alt="EXITING FROM THE HAUNTEDSHELL CLI TERMINAL" width="800" height="131"&gt;&lt;/a&gt;&lt;/p&gt;

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

&lt;p&gt;You can explore, fork, and inspect the entire open-source repository on GitHub: &lt;a href="https://github.com/HarshL14/haunted-shell-ai" rel="noopener noreferrer"&gt;https://github.com/HarshL14/haunted-shell-ai&lt;/a&gt;&lt;/p&gt;

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

&lt;p&gt;HauntedShell is built from scratch using pure Python, keeping dependencies minimal and lightning-fast:&lt;/p&gt;

&lt;p&gt;Hugging Face Transformers (distilgpt2): Instead of calling paid cloud APIs, the script downloads and runs an open-weight transformer model directly on my local CPU/GPU.&lt;/p&gt;

&lt;p&gt;Colorama: Handles terminal text styling, adding warning reds, mystical magentas, and eerie ghost-greens.&lt;/p&gt;

&lt;p&gt;Typewriter Effect Engine: A custom-built asynchronous printing loop (sys.stdout.write with micro-delays) that makes the AI's responses look like they are being typed out live by a restless spirit.&lt;/p&gt;

&lt;p&gt;Interactive Terminal Loop: A continuous while True loop that catches your standard inputs, formats them as seed prompts for the model, and prints out generated poetic commentary.&lt;/p&gt;

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

&lt;p&gt;Building this project made it crystal clear why open-source AI is so powerful for creative developer tools:&lt;/p&gt;

&lt;p&gt;True Offline Functionality: Because the model weights are stored locally on my laptop, the haunted terminal works completely without internet access. It works in the dark, during power outages, or deep in the middle of the night.&lt;/p&gt;

&lt;p&gt;Absolute Privacy: When you are typing commands or venting into a local tool at 2 AM, your thoughts and inputs never leave your hard drive. No external server logs your terminal activity.&lt;/p&gt;

&lt;p&gt;Zero Cost &amp;amp; Freedom: No subscription paywalls, no API token limits, and no credit card required. Open-weight models give developers total creative ownership.&lt;/p&gt;

&lt;h2&gt;
  
  
  The Spirit of It
&lt;/h2&gt;

&lt;p&gt;There is a unique kind of magic in late-night coding. When the house is quiet and the only light comes from your monitor, your computer starts to feel less like a collection of silicon chips and more like an independent entity.&lt;/p&gt;

&lt;p&gt;Building HauntedShell captured that exact nostalgic feeling—blending retro text-based interfaces with modern AI text generation. Watching the terminal talk back with unexpected commentary for the first time genuinely gave me chills!&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
      <category>ai</category>
      <category>python</category>
    </item>
    <item>
      <title>LazyNotes AI</title>
      <dc:creator>Harsh Lone</dc:creator>
      <pubDate>Sun, 25 Oct 2026 19:07:30 +0000</pubDate>
      <link>https://dev.to/harshlone14/lazynotes-ai-225d</link>
      <guid>https://dev.to/harshlone14/lazynotes-ai-225d</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week3-2026-10-19"&gt;Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;I built LazyNotes AI, a lightweight Python tool powered by an open-weight model that automatically reads messy, unorganized notes, meeting logs, or raw text dumps and condenses them into clear summaries.&lt;/p&gt;

&lt;p&gt;Between managing daily tasks, exam prep, and constant notifications, I constantly catch myself putting off reading long logs or sorting through unstructured quick notes. LazyNotes channels inner "sloth energy" by automating that tedious reading process so I can stay lazy while my local machine handles the heavy lifting in the background.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;Here is a quick snapshot of LazyNotes AI running locally in terminal, processing raw text and saving a clean summary to disk:&lt;br&gt;
Initializing LazyNotes Sloth AI (Running Locally)...&lt;/p&gt;

&lt;p&gt;Reading messy text from file...&lt;br&gt;
Today was super chaotic. I had 5 different meetings about the database migration, &lt;br&gt;
forgot to buy groceries, need to finish my hacktoberfest submission by tonight, and my &lt;br&gt;
laptop battery keeps draining fast because of background apps...&lt;/p&gt;

&lt;p&gt;Sloth AI is processing locally...&lt;/p&gt;

&lt;h2&gt;
  
  
  Lazy Summary Result:
&lt;/h2&gt;

&lt;h2&gt;
  
  
  I had 5 different meetings about the database migration, need to finish my hacktoberfest submission by tonight, and remember to check pull requests.
&lt;/h2&gt;

&lt;p&gt;Saved clean summary to summary_output.txt successfully!&lt;/p&gt;

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

&lt;p&gt;Check out the full open-source code on GitHub:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://github.com/HarshL14/lazy-notes-ai" rel="noopener noreferrer"&gt;https://github.com/HarshL14/lazy-notes-ai&lt;/a&gt;&lt;/p&gt;

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

&lt;p&gt;LazyNotes AI is built using Python, Hugging Face Transformers, and PyTorch.&lt;/p&gt;

&lt;p&gt;Instead of relying on cloud APIs, it loads an open-weight model (sshleifer/distilbart-cnn-12-6) directly onto the local machine. The script reads input text from a local .txt file, runs local model inference to generate a summary, and outputs the result into both the terminal and a clean text file automatically.&lt;/p&gt;

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

&lt;p&gt;Open innovation was crucial for this project. Running an open-weight model locally provided three main advantages over closed APIs:&lt;/p&gt;

&lt;p&gt;Complete Data Privacy: Personal daily notes, work logs, and raw thoughts never leave my computer or hit an external server.&lt;/p&gt;

&lt;p&gt;Zero Operational Cost: There are no API keys to generate, no pay-per-token usage fees, and no monthly subscriptions.&lt;/p&gt;

&lt;p&gt;100% Offline Functionality: Once downloaded, the model runs entirely offline without needing an active internet connection.&lt;/p&gt;

&lt;p&gt;Open-source AI made it possible to create a reliable, private tool that works forever for free.&lt;/p&gt;

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

&lt;p&gt;LazyNotes saved me around 15–20 minutes today that I would have otherwise spent manually reading through unorganized notes and writing down action items. I can now drop raw text into a file, run a single command, and get straight to relaxing!&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>python</category>
      <category>hf26challenge</category>
      <category>ai</category>
    </item>
  </channel>
</rss>
