SatyaTrail — News Authenticity & Trail Explorer

A full-stack news verification platform that traces the origin and propagation of news stories through a network of sources, using AI-powered multi-agent verification, blockchain storage, and an intuitive visualization interface.

🌟 Features

Frontend Features

• Tile-based Feed: Browse news articles with veracity badges and dynamic images
• Source Network Visualization: Interactive 3D graph showing how stories propagate from origin through amplifiers, debunkers, and commentators
• Multi-Agent Verification: ChatGPT-style claim checker with real-time agent responses
• Agent Rankings: Leaderboard showing agent reputation scores, accuracy rates, and credibility metrics
• Source Reputation System: Each source scored based on historical accuracy and credibility
• Trail Explorer: Visualize the complete trail of how a story spread across sources
• Recent Verifications Sidebar: Quick access to previously verified claims
• Neo-Brutalist Design: Bold, high-contrast UI with newspaper-inspired styling

Backend Features

• Multi-Agent Orchestration: TOI, NDTV, India Times, and Generic agents analyze claims
• Source Graph Tracing: Maps how news propagates across sources with role classification
• Blockchain Storage: Immutable verification records on Polygon/Solana
• Reputation System: ELO-inspired algorithm tracking agent performance over time
• Tavily Search Integration: Web search and content extraction for evidence gathering
• OpenAI GPT Integration: AI-powered analysis and reasoning
• Bot Integrations: Telegram, Twitter/X, and WhatsApp bots for public verification

Extension Features

• Full Page Verification: Analyze entire webpages for misinformation
• Quick Check: Verify selected text instantly
• Detailed Results: Claim-by-claim verdicts with evidence
• Evidence Sources: View supporting evidence from trusted sources

🛠️ Tech Stack

Frontend

• Framework: Vite + React.js (JavaScript)
• Styling: TailwindCSS with Neo-Brutalism theme
• UI Kit: Radix UI primitives + Lucide icons
• State: Zustand for global stores + URL search params
• Forms: react-hook-form + zod
• 3D Visualization: Three.js + React Three Fiber
• Animations: Framer Motion
• Notifications: sonner

Backend

• Runtime: Node.js 18+
• Framework: Express.js
• Database: MongoDB with Mongoose
• AI: OpenAI GPT-4o
• Search: Tavily API
• Blockchain: Polygon (Mumbai testnet) / Solana (devnet)
• Bots: Telegraf, Twitter API v2, Twilio

Blockchain

• Network: Polygon Mumbai / Solana Devnet
• Smart Contracts: Solidity (Hardhat)
• Storage: Immutable verification hashes on-chain

🏗️ Architecture

System Architecture Diagram

graph TB
    subgraph "Client Layer"
        FE[Frontend<br/>React + Vite]
        EXT[Chrome Extension<br/>Content Script]
        BOT_TG[Telegram Bot]
        BOT_TW[Twitter Bot]
        BOT_WA[WhatsApp Bot]
    end

    subgraph "API Gateway"
        API[Express.js API Server<br/>Port 3001]
    end

    subgraph "Backend Services"
        ORCH[Orchestrator<br/>Multi-Agent Coordinator]
        AG_TOI[TOI Agent]
        AG_NDTV[NDTV Agent]
        AG_IT[India Times Agent]
        AG_GEN[Generic Agent]
        REP[Reputation System<br/>ELO Algorithm]
        GRAPH[Graph Service<br/>Source Network Builder]
    end

    subgraph "External APIs"
        OPENAI[OpenAI GPT-4o<br/>AI Analysis]
        TAVILY[Tavily API<br/>Web Search]
    end

    subgraph "Data Layer"
        MONGO[(MongoDB<br/>Verifications<br/>Reputations<br/>Source Graphs)]
    end

    subgraph "Blockchain"
        POLYGON[Polygon Mumbai<br/>Smart Contract]
        SOLANA[Solana Devnet<br/>Alternative]
    end

    FE -->|HTTP/REST| API
    EXT -->|HTTP/REST| API
    BOT_TG -->|Webhooks| API
    BOT_TW -->|Webhooks| API
    BOT_WA -->|Webhooks| API

    API -->|Verify Request| ORCH
    API -->|Fetch Data| MONGO
    API -->|Store Hash| POLYGON
    API -->|Reputation Queries| REP

    ORCH -->|Delegate| AG_TOI
    ORCH -->|Delegate| AG_NDTV
    ORCH -->|Delegate| AG_IT
    ORCH -->|Delegate| AG_GEN

    AG_TOI -->|AI Analysis| OPENAI
    AG_NDTV -->|AI Analysis| OPENAI
    AG_IT -->|AI Analysis| OPENAI
    AG_GEN -->|AI Analysis| OPENAI

    AG_TOI -->|Evidence Search| TAVILY
    AG_NDTV -->|Evidence Search| TAVILY
    AG_IT -->|Evidence Search| TAVILY
    AG_GEN -->|Evidence Search| TAVILY

    ORCH -->|Build Graph| GRAPH
    GRAPH -->|Store| MONGO

    ORCH -->|Update Scores| REP
    REP -->|Store| MONGO

    ORCH -->|Store Results| MONGO
    ORCH -->|Store Hash| POLYGON

    style FE fill:#6366f1,stroke:#4f46e5,color:#fff
    style EXT fill:#6366f1,stroke:#4f46e5,color:#fff
    style API fill:#10b981,stroke:#059669,color:#fff
    style ORCH fill:#f59e0b,stroke:#d97706,color:#fff
    style MONGO fill:#47a248,stroke:#2e7d32,color:#fff
    style OPENAI fill:#8b5cf6,stroke:#7c3aed,color:#fff
    style TAVILY fill:#ec4899,stroke:#db2777,color:#fff
    style POLYGON fill:#8247e5,stroke:#6b21a8,color:#fff

Loading

🚀 Getting Started

Prerequisites

• Node.js 18 or higher
• MongoDB (local or cloud)
• API keys for:
  • OpenAI
  • Tavily
  • (Optional) Telegram bot token, Twitter API credentials

Frontend Setup

# Install dependencies
npm install

# Run development server
npm run dev

# Build for production
npm run build

# Preview production build
npm run preview

The frontend will run on http://localhost:5173 (or the next available port).

Backend Setup

# Navigate to backend directory
cd backend

# Install dependencies
npm install

# Configure environment
cp .env.example .env
# Edit .env with your API keys

# Start MongoDB (if using Docker)
docker-compose up -d mongo

# Run development server (with auto-reload)
npm run dev

# Or run production server
npm start

The backend will run on http://localhost:3001 (or configured port).

Extension Setup

1. Open Chrome and navigate to chrome://extensions/
2. Enable Developer mode (toggle in top right)
3. Click Load unpacked
4. Select the extension folder from this project
5. The SatyaTrail icon should appear in your toolbar

Verification Flow Diagram

sequenceDiagram
    participant User
    participant Frontend
    participant API
    participant Orchestrator
    participant Agents
    participant OpenAI
    participant Tavily
    participant GraphService
    participant Reputation
    participant MongoDB
    participant Blockchain

    User->>Frontend: Submit Claim/URL
    Frontend->>API: POST /api/v1/verify
    API->>Orchestrator: Initiate Verification
    
    par Parallel Agent Processing
        Orchestrator->>Agents: Delegate to TOI Agent
        Orchestrator->>Agents: Delegate to NDTV Agent
        Orchestrator->>Agents: Delegate to India Times Agent
        Orchestrator->>Agents: Delegate to Generic Agent
    end

    loop For Each Agent
        Agents->>Tavily: Search for Evidence
        Tavily-->>Agents: Return Search Results
        Agents->>OpenAI: Analyze Claim + Evidence
        OpenAI-->>Agents: Return Analysis & Verdict
        Agents-->>Orchestrator: Agent Report
    end

    Orchestrator->>Orchestrator: Aggregate Verdicts
    Orchestrator->>GraphService: Build Source Graph
    GraphService->>GraphService: Classify Roles<br/>(Origin/Amplifier/Debunker/Commentary)
    GraphService-->>Orchestrator: Source Graph
    
    Orchestrator->>Reputation: Update Agent Scores
    Reputation->>Reputation: Calculate ELO Updates
    Reputation->>MongoDB: Store Updated Reputations
    
    Orchestrator->>MongoDB: Store Verification Result
    Orchestrator->>Blockchain: Store Graph Hash
    
    Blockchain-->>Orchestrator: Transaction Hash
    Orchestrator-->>API: Verification Complete
    API-->>Frontend: Return Results
    Frontend-->>User: Display Verification

Loading

🎯 Source Network Role Classification

The system classifies sources in the trail network into four roles:

• Origin: The original source where the story first appeared (earliest timestamp)
• Amplifier: High-credibility sources that spread the story (high reputation + amplifier signals)
• Debunker: Fact-check sources that verify or debunk claims (fact-check domains + debunk signals)
• Commentary: Analysis, opinion pieces, and other commentary on the story

Classification uses:

• Backend-provided roles (if available)
• Timestamp analysis (earliest = origin)
• Domain reputation and fact-check domain lists
• Content signals (keywords, language patterns)
• Credibility scores

📊 Reputation System

The reputation system uses an ELO-inspired algorithm:

• Reputation Score: Base score (1000) adjusted by verification outcomes
• Accuracy Rate: Percentage of verifications where agent agreed with consensus
• Credibility Score: Average credibility score (0-100) across all verifications
• Total Verifications: Count of verifications performed by the agent

Scores are updated after each verification based on:

• Agreement with consensus verdict
• Confidence level of the verification
• Time decay (older verifications have less weight)

🔗 Blockchain Integration

Verification results are stored on-chain for immutability:

• Network: Polygon Mumbai (testnet) / Solana Devnet
• Storage: Graph hash stored on-chain
• Verification: Full verification data stored off-chain, hash on-chain
• Smart Contract: SatyaTrail.sol handles hash storage and retrieval

🤖 Multi-Agent System

The platform uses specialized agents for verification:

• Generic Verification Agent: General-purpose fact-checking
• TOI Agent: Times of India perspective
• NDTV Agent: NDTV perspective
• India Times Agent: India Times perspective

Each agent:

• Analyzes claims independently
• Provides reasoning and evidence
• Assigns credibility scores
• Contributes to consensus verdict