JSO Agency Trust & Transparency Agent

An AI-powered platform for evaluating and monitoring recruitment agency trustworthiness on the JSO ecosystem.

---

Overview

Users on job platforms cannot easily determine which recruitment agencies are trustworthy — leading to wasted time, fraud exposure, and platform credibility issues. The Agency Trust & Transparency Agent solves this by:

• Computing transparent trust scores (0–100) from reviews, placement success rates, feedback ratings, compliance records, and platform tenure
• Providing a conversational AI interface for querying agency trustworthiness
• Surfacing real-time analytics and trend monitoring across dashboards
• Detecting anomalies (fake reviews, declining agencies) and alerting stakeholders

Prototype

Tech Stack

• Frontend: Next.js 16 + React + TypeScript + Tailwind CSS
• Charts: Recharts
• Icons: Lucide React
• Deployment-ready: Vercel-compatible

Running Locally

cd prototype
npm install
npm run dev

Open http://localhost:3000

Features

Feature	Description
Agency Dashboard	Browse agencies sorted by trust score with search, filter, and sort
Trust Score Badges	Visual score circles with tier labels (Excellent/Good/Fair/Poor)
Agency Detail View	Click any agency for full breakdown — score history charts, component analysis, reviews
Compare Agencies	Side-by-side radar chart comparison of up to 3 agencies with detailed metrics table
Admin Reports	Platform-wide analytics — flagged agencies, trust rankings, trend lines, compliance overview
AI Chat Agent	Conversational interface to ask about agency trust (bottom-right chat button)
Trust Analytics	Pie charts and bar charts showing trust distribution and placement rates
Score Explanation	Transparent algorithm display showing all weights and factors

API Endpoints

Endpoint	Method	Purpose
/api/agencies	GET	List all agencies with optional filters
/api/agencies/[id]	GET	Get agency details + trust score breakdown
/api/agent/chat	POST	Send natural language query to the trust agent

Architecture

Client (Next.js + React)
  → API Layer (Next.js API Routes)
    → Agent Orchestration (Trust Score Calculator + Chat Agent)
      → Data Layer (Mock data / Supabase in production)
        → AI Layer (Claude API in production)

Trust Score Algorithm

Trust Score = (0.30 × Reviews) + (0.25 × Placements) + (0.20 × Feedback) 
            + (0.15 × Compliance) + (0.10 × Tenure)

Tier	Score Range
Excellent	80–100
Good	60–79
Fair	40–59
Poor	0–39

Project Structure

JSO-Agent/
├── README.md                # This file
└── prototype/               # Next.js application
    └── src/
        ├── app/
        │   ├── page.tsx             # Main dashboard page
        │   ├── components/          # UI components
        │   │   ├── AdminReports.tsx   # Admin trust reports & flagged agencies
        │   │   ├── AgencyCard.tsx     # Agency listing card
        │   │   ├── AgencyDetail.tsx   # Detail modal with charts
        │   │   ├── ChatWidget.tsx     # AI chat interface
        │   │   ├── CompareView.tsx    # Side-by-side agency comparison
        │   │   ├── DashboardStats.tsx # Stats overview
        │   │   └── TrustDistribution.tsx # Analytics charts
        │   └── api/
        │       ├── agencies/         # Agency API routes
        │       └── agent/chat/       # Chat agent API
        └── lib/
            └── data.ts              # Data models, mock data, scoring logic

Deployment

Deploy to Vercel with:

• Root Directory: prototype
• Framework: Next.js (auto-detected)
• Build Command: npm run build
• No environment variables required