Skip to main content

📋 Product Backlog

The development roadmap for ChildSafeNet.
This backlog organizes system features into structured phases covering core platform, AI detection, browser integration, and deployment readiness.

RoadmapPhased DevelopmentRelease Ready

🏗 Architecture

🧠 AI Training

🔐 Security

1. Overview​

The Product Backlog defines the main development tasks for the ChildSafeNet platform.

The backlog is structured into five development phases, each focusing on a specific part of the system:

  • Core system functionality
  • AI detection and browser integration
  • Admin tools and training pipelines
  • Documentation and release preparation
  • DevOps and security hardening

This structure ensures clear milestones, modular development, and stable releases.

2. Development Phases​

PhaseFocus
Phase 1Core system (auth, settings, logs)
Phase 2AI detection and browser extension
Phase 3Admin tools and dataset training
Phase 4Documentation and release
Phase 5DevOps and security hardening

3. Core System (Phase 1)​

1
Auth

Authentication & Authorization

Core security layer for user access.

Key tasks:

  • Implement JWT authentication
  • Support Parent and Admin roles
  • Secure password hashing
  • Protect API routes

2
Settings

Parent Settings

Allow parents to configure protection policies.

Key tasks:

  • Strict / Balanced / Relaxed modes
  • Whitelist and blacklist
  • Protection toggle
  • Settings persistence

3
Logs

Scan & Activity Logs

Record scan activity and provide transparency.

Key tasks:

  • /api/scan endpoint
  • ScanLog database table
  • Paginated logs API
  • Logs dashboard UI

4. AI & Browser Integration (Phase 2)​

1
AI

Hybrid AI Detection Engine

Machine learning system for detecting unsafe URLs.

Key tasks:

  • Extract 1000+ handcrafted URL features
  • Build TF-IDF NLP pipeline
  • Combine scores via aggregation logic
  • Configure detection thresholds
  • Serialize trained models

2
Extension

Browser Extension (MV3)

Real-time protection within the browser.

Key tasks:

  • Navigation listener
  • Scan request to API
  • Pairing token flow
  • Block page (block.html)
  • Mode synchronization

5. Admin & Training (Phase 3)​

1
Dataset

Dataset Review

Admin interface to manage training samples.

Key tasks:

  • UrlDataset schema
  • Approve / Reject samples
  • Dataset review UI
  • Dataset export (CSV)

2
Training

Periodic Model Training

Scheduled retraining pipeline.

Key tasks:

  • Background training job
  • Merge baseline + approved dataset
  • Compute evaluation metrics
  • Register model versions
  • Safe activation + rollback

6. Documentation & Release (Phase 4)​

Key tasks:

  • Setup Docusaurus documentation site
  • Write architecture documentation
  • Create API reference
  • Provide Parent/Admin user guides
  • Prepare release changelog

7. DevOps & Security (Phase 5)​

Key tasks:

  • GitHub Actions CI/CD pipeline
  • Dockerfile + Docker Compose
  • Environment configuration
  • HTTPS enforcement
  • Security headers
  • Rate limiting
  • Audit logging

8. Definition of Done​

A backlog task is complete when:

  • Code implementation finished
  • Tests pass successfully
  • Security checks pass
  • Code reviewed and merged
  • Documentation updated
  • No critical regression introduced

9. Long-Term Vision (Post-Release)​

  • Model drift detection
  • Online evaluation dashboard
  • Multi-device management
  • Explainability visualization (feature importance UI)
  • Advanced threat intelligence integration