LO-MA-BAN

Inspiration

The idea for ScoutNet came from a simple observation.

A team member's younger cousin clicked a phishing link. The browser blocked it with a stark "Access Denied" message. The child's reaction wasn't relief—it was frustration.

"Why can't I go there? What did I do wrong?"

We researched existing solutions and found a troubling pattern:

Approach	What Happens	The Problem
Direct Blocking	Browser blocks the site	Child doesn't learn why
Parental Controls	Parents monitor activity	Child feels distrusted
School Lessons	Teach safety in class	Lessons feel disconnected

Children were being acted upon, not engaged with.

We asked ourselves: What if we could turn every blocked site into a learning moment? What if instead of saying "no" we asked "why do you think this might be unsafe?"

ScoutNet was born—not as another blocking tool, but as a guided safety companion.

What It Does

ScoutNet is a Chrome extension that transforms online safety from passive blocking into active learning.

The Core Flow

1. Trigger User clicks a link or enters a URL, triggering ScoutNet's background service worker.

2. Security Scanning (4 APIs in Parallel) The URL is immediately sent to four security intelligence APIs — VirusTotal, URLhaus, PhishTank, and Google Safe Browsing — running in parallel for maximum speed.

3. Risk Routing

Path A — Phishing Detected: If any security API flags the URL as malicious, ScoutNet calls Featherless AI for deep phishing analysis. The AI generates a child-friendly security report including predicted intended URLs, alternative safe websites, evidence cards, an interactive quiz, and safety tips.
Path B — No Security Risk: The URL is passed to Exa AI to fetch the actual page content. If the fetch fails, ScoutNet falls back to searching related content (e.g. from Reddit or other sources).
- Featherless AI then classifies whether the content is child-friendly.
- Child-friendly → User enters the site directly.
- Not child-friendly → A content risk report is generated (same structure as phishing reports).

4. URL Security Analysis Report The report is presented as an interactive overlay containing:

Verified safe URL information & executive summary
Recommended actions
FAQ & interactive support (quiz)

5. User Decision — "Still want to access?"

No → Break. The user leaves safely.
Yes → The user must provide a reason for why they still want to enter.

6. Second-Stage Persuasion The user's reason is sent to Featherless AI, which responds with empathetic persuasion and education — including behavior consequence warnings, empathy notes, and encouraging messages.

7. Final Outcome

If the user accepts the guidance → Break (leave).
If the user insists on accessing → Enter (with persistent risk markers).

How We Built It

Tech Stack

Layer	Technology
Frontend	React + TypeScript + Vite
Extension	Chrome Manifest V3
Backend	FastAPI (Python, async)
Security APIs	VirusTotal, URLhaus, PhishTank, Google Safe Browsing
Content Analysis	Exa AI
Generative AI	Qwen via Featherless AI
Storage	Chrome Storage API (local only)

Challenges We Ran Into

1. First-Time Chrome Extension Development

Building our first Chrome extension came with unexpected hurdles. Service workers can't access the DOM directly, and content scripts can't use Chrome APIs

Accomplishments that we're proud of

1. Successfully Packaged as a Chrome Extension

We built, packaged, and deployed a fully functional Chrome extension from scratch—our first time working with Chrome's extension architecture. The extension runs continuously in the background, actively monitoring every navigation in real-time.

What we learned

Building ScoutNet taught us how to combine multiple AI technologies into a single, cohesive product. We integrated security APIs (VirusTotal, URLhaus, PhishTank, Google Safe Browsing) with content analysis (Exa AI) and generative AI (Qwen via Featherless) to create a complete safety solution.

We also learned the entire Chrome Extension development lifecycle—from initial concept to packaging, testing, and preparing for Chrome Web Store release. Understanding Manifest V3, service workers, content scripts, and cross-context messaging was a steep but rewarding learning curve.

Most importantly, we learned that children respond to guidance, not commands. When treated as partners in safety rather than subjects of protection, they actively engage and want to learn.

What's next for ScoutNet

Phase 1: MVP Launch & Community Feedback

Our immediate next step is to release the MVP for real users. This will allow us to:

Collect user reports of suspicious websites to expand our threat database Gather feedback on guided questions—are they engaging? Age-appropriate? Effective? Identify edge cases where our AI needs improvement Refine the learning experience based on how children actually interact with the system

❌ Before ScoutNet	✅ After ScoutNet
Close the site quickly	Understand why it's problematic
Feel scared or confused	Feel empowered and informed
Avoid thinking about it	Self-reflect on online safety
Passive reaction	Active awareness

We want children to develop 自主權 (autonomy) and 自我意識 (self-awareness)—not just memorize rules, but truly understand website safety and recognize risks on their own.

Phase 2: Enhanced Learning & Safety Features

ScoutNet will evolve beyond simple warnings into a comprehensive safety learning companion:

Advanced Guided Learning

Refine question generation based on real user interactions Adapt difficulty to each child's learning pace Track progress and suggest new safety topics Make learning feel like exploration, not instruction

Expanded Risk Detection

Continuously update threat database with user-reported sites Improve AI recognition of subtle phishing patterns Detect more types of harmful content (violence, cyberbullying, etc.)

Safety Button Enhancement

Phase 3: Multi-Tiered Business Model

We'll introduce sustainable funding while keeping core features free:

🆓 Free Version (Always Free)

Core safety detection Basic guided questions Community support 👪 Family Plan (Premium)

Multiple child profiles Detailed learning progress reports Custom safety rules Priority support 🏫 School/Classroom Plan

Teacher dashboard for class monitoring Anonymous statistics for educators Curriculum-aligned safety lessons Special reporting for harmful content (bullying, violence, explicit material) When students encounter harmful content, teachers can receive automatic notifications via screenshot and engage with AI to understand appropriate responses.

Phase 4: Long-Term Vision

ScoutNet's ultimate goal is to transform how children learn online safety—from passive recipients of protection to active participants with a genuine understanding.

We envision a future where:

Children develop critical thinking skills about every site they visit Online safety becomes a habit, not homework Parents and teachers become partners in guidance, not monitors Every child, regardless of background, has access to safety education "This isn't just about blocking bad sites. It's about raising a generation that naturally questions, understands, and protects themselves online."