# 👻 GhostShell AI — The Terminal That Talks Back

Srinithi — Wed, 28 Oct 2026 09:57:07 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

What I Built

GhostShell is a spooky, AI-powered developer companion that brings a little supernatural personality to learning terminal commands, exploring programming concepts, and creating eerie stories.

Inspired by the theme Ghost in the Machine, I wanted to build something that combines the familiar experience of a developer terminal with the mysterious personality of an AI that seems to have a life of its own.

GhostShell has three interactive modes:

🔮 Command Oracle: Learn terminal commands through explanations, practical examples, and safety guidance.
💻 Code Séance: Explore programming concepts and get beginner-friendly coding help.
👻 Haunted Storyteller: Generate eerie stories about haunted computers, mysterious code, and supernatural technology.

Rather than relying on predefined responses, GhostShell uses real AI-generated responses, making each interaction an opportunity to learn or discover something unexpected.

The goal was to make learning technology feel less intimidating, more interactive, and a little spooky.

Demo

👻 Try GhostShell live:

https://srinithivijayakumar82-cloud.github.io/ghostshell-ai/

Choose a mode, enter a prompt, and summon a response.

Try asking the Command Oracle what pwd does, asking Code Séance to explain a Python loop, or asking the Haunted Storyteller for a story about a computer that starts receiving messages from the future.

Code

🔗 GitHub Repository:

https://github.com/srinithivijayakumar82-cloud/ghostshell-ai

The project is built with HTML, CSS, and JavaScript, with a Cloudflare Worker connecting the frontend to Groq's AI inference service.

The repository contains the source code and project documentation.

How I Built It

I built GhostShell using a lightweight web stack so I could focus on the interactive experience without introducing an unnecessarily complicated frontend framework.

Technologies used:

HTML5 — Structure of the application.
CSS3 — Dark, terminal-inspired visuals, styling, and responsive layout.
JavaScript — User interactions, mode selection, and API requests.
Groq API — AI-generated responses.
Cloudflare Workers — Backend request handling and secure server-side API-key management.
GitHub Pages — Hosting the frontend.
Marked.js and DOMPurify — Rendering Markdown responses while sanitizing generated HTML.

The request flow is:

Browser → Cloudflare Worker → Groq API → Cloudflare Worker → Browser

When someone submits a prompt, the frontend sends the selected mode and prompt to the Worker. The Worker forwards the request to Groq and returns the generated response, which GhostShell displays in its terminal-inspired interface.

One important lesson was that connecting an AI model is only part of the work. I also had to deal with deployment, cross-origin requests (CORS), frontend-backend communication, and keeping the API key out of public browser code.

This project helped me understand how separate web technologies can work together to create an AI-powered application.

Why Does Open Innovation Matter?

Open innovation makes it possible for developers to experiment, learn from existing tools, and turn ideas into working projects without having to build every component from scratch.

For GhostShell, accessible AI inference through Groq made it possible to add dynamic AI responses to a lightweight web application. Open-source frontend technologies and publicly available development tools helped me build and deploy the experience while learning along the way.

It also means other developers can explore the code, suggest improvements, adapt the interface, or build their own creative experiences on top of similar ideas.

For me, open innovation is about lowering the barrier between “I have an idea” and “I built something people can try.”

The Spirit of It

The idea behind GhostShell is simple: what if your terminal didn't just execute instructions, but also explained them, helped you understand code, and occasionally told you a story that made you question whether you should keep your computer switched on?

The spooky theme isn't just decoration. It gives each mode a personality and turns an ordinary interaction with an AI assistant into something more playful.

Behind the ghostly interface is a practical goal: helping people feel more comfortable exploring technology.

Sometimes the machine doesn't need to be haunted. It just needs a good personality. 👻

Built with curiosity, code, and a little supernatural assistance. 👻

SellCraft

Srinithi — Mon, 19 Oct 2026 06:58:06 +0000

*This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle*SellCraft
SellCraft is an AI-powered product listing generator designed to help
small businesses, online sellers, and side hustlers create product copy
more quickly.
What it does
SellCraft provides a simple web interface for entering product details
and generating marketing copy, such as:

A product title
A product description
A social media caption The goal is to reduce the time sellers spend writing product listings and make it easier to prepare copy for online stores and social media. Project status The frontend and Express server have been created. The current server implementation is configured to call the Hugging Face Inference API using the Qwen/Qwen2.5-7B-Instruct model.

Important: The hosted inference request currently cannot be relied
on because the Hugging Face account returned a 402 Payment Required
response indicating that no inference credits remained. Local inference
with Ollama is planned, but should be described as working only after
the model has downloaded successfully and the app has been connected to
it and tested.

Built with

HTML, CSS, and JavaScript for the frontend
Node.js and Express for the backend
Hugging Face Inference API integration in the current server implementation
Qwen2.5 model configuration Run locally Prerequisites
Node.js and npm
A Hugging Face access token with the required inference permissions, if using the current hosted-inference implementation
Clone the repository git clone (https://github.com/srinithivijayakumar82-cloud/Sellcraft) cd SellCraft
Install dependencies npm install
Configure your environment Create a .env file in the project root (the same folder as server.js) and add your own token: HF_TOKEN=your_hugging_face_token_here Never commit .env or share your token publicly. The .gitignore file should include: node_modules/ .env
Start the server npm start Then open http://localhost:3000 in your browser.

Security

Keep API tokens in environment variables, not in source code.
Do not upload .env to GitHub.
Do not upload node_modules; dependencies can be installed with npm install.
If a token has ever been published publicly, revoke it and create a replacement. Open innovation SellCraft is an open-source project intended to explore how AI can help small sellers create useful product copy. Open models and local inference can offer developers more control over model selection, experimentation, and where inference runs. The local-inference path is still a work in progress and should not be presented as completed until it has been integrated and tested. Current limitations and next steps
Restore working AI generation by using a provider with available inference access or by integrating and testing a local Ollama model.
Add validation and clearer error handling for model/provider failures.
Test the full product-listing flow from the browser through the backend.
Add a demo link or video once a working demo is available. Contributing Issues and improvements are welcome. Please avoid committing secrets, environment files, or generated dependency folders.

``

What I Built

Demo

Code

How I Built It

srinithivijayakumar82-cloud / Sellcraft

SellCraft SellCraft is an AI-powered product listing generator designed to help small businesses, online sellers, and side hustlers create product copy more quickly What it does SellCraft provides a simple web interface for entering product details and generating marketing copy, such as:

A product title
A product description
A social media caption The goal is to reduce the time sellers spend writing product listings and make it easier to prepare copy for online stores and social media Project status The frontend and Express server have been created. The current server implementation is configured to call the Hugging Face Inference API using the Qwen/Qwen2.5-7B-Instruct model.

Important: The hosted inference request currently cannot be relied on because the Hugging Face account returned a 402 Payment Required response indicating that no inference credits remained. Local inference with Ollama is planned, but should be described as working only after the model has downloaded successfully…

View on GitHub

DEV Community: Srinithi

# 👻 GhostShell AI — The Terminal That Talks Back

What I Built

Demo

Code

How I Built It

Why Does Open Innovation Matter?

The Spirit of It

SellCraft

What I Built

Demo

Code

How I Built It

srinithivijayakumar82-cloud / Sellcraft