A cryptocurrency investigation platform that analyzes crypto projects for legitimacy, scams, and security risks. Built with React, TypeScript, Vite, and Node.js.

• Crypto Investigation: Search and investigate cryptocurrencies by name or symbol
• 3-Step Analysis Pipeline:
  • Planner: Generates crypto-specific search queries
  • Eyes: Scrapes web data using Tavily API (news, security audits, exchange hacks)
  • Brain: Analyzes results with Google Gemini AI to generate risk scores
• Legitimacy Scoring: Automated risk assessment (0-100) with verdicts: "LIKELY RISKY", "SORT OF RISKY", or "SAFE"
• Comparison Tool: Compare multiple cryptocurrencies side-by-side
• Share Investigations: Generate shareable links for investigation results
• Comment Threads: Add comments and discussions on specific findings
• Real-time Agent Logs: Track investigation progress in real-time
• Evidence Board: Visual display of investigation findings organized by category

• React 18 + TypeScript
• Vite (build tool)
• Tailwind CSS (styling)
• Radix UI + shadcn/ui (UI components)
• React Router (routing)
• TanStack Query (state management)

• Node.js + Express (API server)
• Tavily API (web scraping/search)
• Google Gemini AI (analysis and scoring)
• CORS (cross-origin support)
• dotenv (environment variables)

• Node.js 18+
• npm, yarn, or pnpm
• API Keys:
  • Tavily API key
  • Google Gemini API key

1. Clone the repository:

git clone <repository-url>
cd deep-detective-clean

2. Install dependencies:

npm install

3. Create a .env file in the root directory:

TAVILY_API_KEY=your_tavily_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here

npm run dev:all

This starts both servers concurrently:

• Frontend: http://localhost:5173
• Backend API: http://localhost:3001

Terminal 1 - Backend:

npm run server

Terminal 2 - Frontend:

npm run dev

deep-detective-clean/
├── server.js              # Express backend server
├── .env                   # Environment variables (create this)
├── src/
│   ├── components/       # React components
│   │   ├── ui/          # shadcn/ui components
│   │   ├── AgentLog.tsx
│   │   ├── CaseInput.tsx
│   │   ├── FindingComments.tsx
│   │   ├── ManilaFolder.tsx
│   │   ├── ShareInvestigation.tsx
│   │   └── VerdictPanel.tsx
│   ├── pages/            # Page components
│   │   ├── Index.tsx     # Home/input page
│   │   ├── Results.tsx   # Investigation results
│   │   ├── Comparison.tsx # Comparison view
│   │   └── Shared.tsx    # Shared investigation view
│   ├── lib/              # Utilities and API clients
│   │   ├── api/
│   │   │   └── investigate.ts
│   │   ├── comments.ts
│   │   ├── comparison.ts
│   │   └── sharing.ts
│   ├── types/            # TypeScript definitions
│   └── assets/           # Static assets
├── public/               # Public assets
└── package.json

Investigate a cryptocurrency.

Request Body:

{
  "target": "Bitcoin"
}

Response:

{
  "plan": ["query1", "query2", "query3"],
  "logs": [
    {
      "query": "Bitcoin crypto scam fraud rug pull",
      "data": [
        {
          "title": "Article Title",
          "content": "Article content...",
          "url": "https://..."
        }
      ]
    }
  ],
  "report": {
    "score": 85,
    "flags": ["flag1", "flag2"],
    "verdict": "Summary verdict..."
  }
}

npm run build

npm run preview

npm test          # Run tests once
npm run test:watch # Run tests in watch mode

npm run lint

1. User Input: Enter a crypto name or symbol (e.g., "Bitcoin", "BTC", "Ethereum")
2. Planner: Generates 3 crypto-specific search queries:
   • Crypto scam/fraud/rug pull searches
   • Security audit/vulnerability searches
   • Exchange hack/exploit searches
3. Eyes: Executes searches in parallel using Tavily API
4. Brain: Aggregates results and sends to Gemini AI for analysis
5. Results: Returns legitimacy score (0-100), red flags, and verdict

The backend includes a demo mode for "FTX" that returns a hardcoded high-risk response for demonstration purposes.

MIT