Reveal your digital shadow before someone else does.
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine.
The Inciting Incident: When Your Digital Shadow Gets Weaponized
It started with a jarring realization.
I had a GitHub account dedicated to personal, experimental projects—things I built in private hours to test my limits. It wasn't linked to my professional resumes, and I hadn't broadcasted it. Yet, a group of highly competitive peers managed to hunt it down, piece together my identity, and quietly map out my unpolished, backend codebases.
When I realized they had been silently combing through my personal digital workspace, I felt completely violated.
The vulnerability wasn't just uncomfortable; it was eye-opening. A username here, an email commit hash there, a pattern of activity somewhere else—harmless traces on their own, but dangerous when correlated. I realized that our digital footprints tell a story we never explicitly gave anyone permission to read.
Instead of retreating, I chose to build a countermeasure. I decided to build a tool to teach people just how much their public presence actually reveals.
Meet CIPHANTOM
What I Built
CIPHANTOM is an open-source OSINT (Open-Source Intelligence) digital footprint awareness engine. It investigates publicly accessible identity infrastructure to show you what a malicious actor or an invasive peer can piece together about you before they do.
I interpreted the Ghost in the Machine theme as this exact phenomenon: the invisible, fragmented version of ourselves that we unknowingly leave scattered across the internet long after we've closed the tab.
With CIPHANTOM, users can input a name and a target username to:
- Identify publicly exposed accounts across platforms (starting with GitHub's public data).
- Examine available profile cross-correlations and leaked metadata.
- Calculate a deterministic, evidence-based risk assessment.
- Simulate highly targeted phishing email based on that public footprint so users understand their vulnerabilities in real-time.
The goal isn't to collect information for the sake of collecting it. It's to make people more aware of what their online presence communicates about them.
Demo
đź”— Try CIPHANTOM
Code
💻 GitHub Repository — CIPHANTOM
Contributions, suggestions, and feedback are welcome!
How I Built It
CIPHANTOM is built around a React frontend and a Node.js/Express backend.
The investigation pipeline follows a simple principle:
Evidence first. Interpretation second.
The backend collects publicly accessible information through available checks, including GitHub's public API. Findings are classified according to their verification status, confidence, and visibility.
A deterministic risk engine evaluates the collected evidence and produces a risk score. This separation is intentional: an AI model should not invent evidence, inflate exposure, or decide a risk score based on speculation.
For the AI component, I integrated Gemma 4 e2b through Ollama, using open-weight AI for local inference.
Rather than handing the entire investigation over to an LLM, I use it as an interpretation layer. Its role is to explain findings, provide context, and generate actionable privacy recommendations while leaving evidence collection and risk calculation to the backend.
The technology stack includes:
- Frontend: React, Vite, JavaScript, CSS
- Backend: Node.js, Express
- OSINT: GitHub public API
- AI: Gemma 4 e2b, Ollama
- Deployment: Render
One important design decision was to distinguish between a confirmed result and an unverified one. A failed lookup is not proof that an account doesn't exist, and missing information should never be presented as a clean bill of digital privacy.
Why Does Open Innovation Matter?
For a project like CIPHANTOM, trust matters.
A tool that evaluates someone's digital exposure should not be a black box that produces impressive-looking scores without explaining where they came from.
Using open-source AI gave me the opportunity to experiment with local inference, understand how an open-weight model fits into an existing backend, and decide exactly what responsibilities the model should and should not have.
It also made the architecture more intentional. The AI doesn't own the investigation. It assists in understanding its results.
Open innovation makes it possible for developers to inspect, modify, experiment with, and build upon the underlying technology instead of depending entirely on proprietary APIs.
I want CIPHANTOM to be something that can evolve through community contributions, better verification methods, and more transparent approaches to digital privacy.
The Spirit of It
We tend to think of our online presence as something we control.
We choose our usernames. We decide what to post. We decide which profiles to make public.
But we rarely think about what happens when those individual pieces of information are viewed together.
A username here. A profile there. A name attached to an account. A public repository. A forgotten connection.
None of these necessarily reveals much on its own. But together, they can create a picture of someone.
The ghost isn't hiding somewhere inside the machine. It's the version of us that we've unknowingly left behind in it.
CIPHANTOM is my attempt to make that ghost visible — before someone else decides to look for it.
Prize Categories
- Gemma: Integrated Gemma 4 e2b through Ollama as the AI interpretation layer for CIPHANTOM's OSINT findings and privacy recommendations.
- Render: Deployed CIPHANTOM as a web application using Render's hosting platform.
Your digital shadow is already out there. The question is: do you know what it reveals?
Top comments (0)