OmniSecure

Solution Overview and Objectives

Overview

OmniSecure is a privacy-first security suite designed to protect end users from accidental data exposure and malicious content during real-time communications and digital interactions.

Objectives

• Real-time Privacy Protection: Detect and blur sensitive information ("help" keywords, PII) on live video streams and screens.
• Phishing Detection: Analyze SMS and email content to score for phishing risk, generate human-readable explanations, and produce PDF reports.
• Rapid Alerting: Use Twilio API to notify guardians or admins when suspicious messages are detected.

Tech Stack

• Frontend:
  • React (Vite) for a responsive web dashboard and attacker/client panels.
  • CSS Modules for theming (light/dark mode).
• Backend Services:
  • Flask for REST API providing SMS simulation and phishing detection.
  • Python libraries: requests, re, urllib.parse for text analysis.
  • Groq API integration for generating synthetic SMS and explanations.
  • Twilio API for alert messaging.
• Real-time Blur Engine:
  • Python + OpenCV + mss for screen capture and rendering.
  • PaddleOCR for text detection.
• Reporting:
  • FPDF for PDF generation of detection reports.
• Storage:
  • Local file system for storing PDF reports.
• Deployment & Dev Tools:
  • Vite for frontend build.
  • pip virtual environment for Python dependencies.
  • Docker (future).

Architecture Diagram

Use draw.io to import the XML below:

<mxfile host=\"app.diagrams.net\">
 <diagram name=\"OmniSecure Architecture\">
  <mxGraphModel dx=\"827\" dy=\"533\" grid=\"1\" gridSize=\"10\">
    <root>
      <mxCell id=\"0\"/>
      <mxCell id=\"1\" parent=\"0\"/>
      <!-- React Frontend -->
      <mxCell id=\"2\" value=\"React Frontend\" style=\"rounded=1;fillColor=#e1f5fe;\" vertex=\"1\" parent=\"1\">
        <mxGeometry x=\"40\" y=\"40\" width=\"160\" height=\"80\" as=\"geometry\"/>
      </mxCell>
      <!-- Flask API -->
      <mxCell id=\"3\" value=\"Flask Backend\" style=\"rounded=1;fillColor=#fff9c4;\" vertex=\"1\" parent=\"1\">
        <mxGeometry x=\"260\" y=\"40\" width=\"160\" height=\"80\" as=\"geometry\"/>
      </mxCell>
      <!-- Privacy Blur Service -->
      <mxCell id=\"4\" value=\"Privacy Blur Service\" style=\"rounded=1;fillColor=#c8e6c9;\" vertex=\"1\" parent=\"1\">
        <mxGeometry x=\"480\" y=\"40\" width=\"160\" height=\"80\" as=\"geometry\"/>
      </mxCell>
      <!-- Groq API -->
      <mxCell id=\"5\" value=\"Groq LLM API\" style=\"rounded=1;fillColor=#ffe0b2;\" vertex=\"1\" parent=\"1\">
        <mxGeometry x=\"260\" y=\"160\" width=\"160\" height=\"60\" as=\"geometry\"/>
      </mxCell>
      <!-- Twilio API -->
      <mxCell id=\"6\" value=\"Twilio SMS API\" style=\"rounded=1;fillColor=#d1c4e9;\" vertex=\"1\" parent=\"1\">
        <mxGeometry x=\"260\" y=\"250\" width=\"160\" height=\"60\" as=\"geometry\"/>
      </mxCell>
      <!-- Storage -->
      <mxCell id=\"7\" value=\"Local FS Reports\" style=\"rounded=1;fillColor=#f8bbd0;\" vertex=\"1\" parent=\"1\">
        <mxGeometry x=\"480\" y=\"160\" width=\"160\" height=\"60\" as=\"geometry\"/>
      </mxCell>
      <!-- Connectors -->
      <mxCell id=\"8\" style=\"edgeStyle=orthogonalEdgeStyle;endArrow=classic;\" edge=\"1\" source=\"2\" target=\"3\" parent=\"1\">
        <mxGeometry as=\"geometry\"/>
      </mxCell>
      <mxCell id=\"9\" style=\"edgeStyle=orthogonalEdgeStyle;endArrow=classic;\" edge=\"1\" source=\"3\" target=\"5\" parent=\"1\">
        <mxGeometry as=\"geometry\"/>
      </mxCell>
      <mxCell id=\"10\" style=\"edgeStyle=orthogonalEdgeStyle;endArrow=classic;\" edge=\"1\" source=\"3\" target=\"6\" parent=\"1\">
        <mxGeometry as=\"geometry\"/>
      </mxCell>
      <mxCell id=\"11\" style=\"edgeStyle=orthogonalEdgeStyle;endArrow=classic;\" edge=\"1\" source=\"3\" target=\"7\" parent=\"1\">
        <mxGeometry as=\"geometry\"/>
      </mxCell>
      <mxCell id=\"12\" style=\"edgeStyle=orthogonalEdgeStyle;endArrow=classic;\" edge=\"1\" source=\"2\" target=\"4\" parent=\"1\">
        <mxGeometry as=\"geometry\"/>
      </mxCell>
    </root>
  </mxGraphModel>
 </diagram>
</mxfile>

Implementation Challenges and Resolutions

• Real-time OCR Performance
  Challenge: PaddleOCR on CPU yields latency.
  Resolution: Implement GPU support and region-of-interest filtering to blur only sensitive keywords.
• False Positives in Phishing Detection
  Challenge: Generic keywords may flag benign messages.
  Resolution: Tuned scoring thresholds and aggregated keywords vs. link-based scoring.
• API Rate Limits
  Challenge: Groq and Twilio quotas constrain live demos.
  Resolution: Added caching for generated SMS and batched alerting.
• Cross-Origin Requests (CORS)
  Challenge: Frontend fetch blocked by CORS.
  Resolution: Enabled flask_cors on all endpoints.

Future Scope and Productionization Plan

• Scalability
  • Containerize services with Docker & Kubernetes.
  • Migrate PDF storage to S3 or database.
• Advanced Privacy Protections
  • Expand blur filters to detect PII (emails, SSNs) using regex/NER.
• Machine Learning Integration
  • Train custom phishing classifier for higher accuracy.
• Monitoring & Logging
  • Integrate Prometheus & Grafana for metrics, ELK stack for logs.
• Authentication & Access Control
  • Add OAuth2 for user login and role-based dashboards.
• Deployment
  • CI/CD pipeline with GitHub Actions, automated tests, and blue/green deploy.

---

README.md should now serve as both documentation and a starting point for onboarding new developers and stakeholders.