Sentin3l AI - Web3 Security Terminal

English | 中文文档 (Chinese)

---

Sentin3l AI is an advanced, real-time Web3 security terminal designed to detect smart contract vulnerabilities across EVM networks. It leverages customizable LLMs (Gemini, DeepSeek, GPT-4o via OpenAI-compatible API) to perform instant code analysis, offering distinct detection modes for different threat models (Rug Pulls, Thefts, and General Audits).

🌟 Why Sentin3l AI?

Unlike standard block explorers or generic AI chats, Sentin3l AI is built specifically for SecOps:

1. 🛡️ Role-Based Intelligence: The AI doesn't just "read code"; it assumes specific personas. In RugPull Mode, it thinks like a paranoid investor looking for honeypots. In Theft Mode, it acts as a black-hat hacker hunting for exploits.
2. 🔒 Local-First Privacy: Zero backend tracking. Your API keys, scan history, and custom RPC URLs are stored exclusively in your browser's LocalStorage. We cannot see your data.
3. ⚡ Real-Time "Matrix" Stream: Don't just audit static addresses. Watch live chain events (Liquidity Adds, Admin Changes, Deployments) and analyze them instantly as they happen.
4. 🧩 Agnostic AI Backend: Plug in any OpenAI-compatible API. You control the intelligence model and the costs.
5. 🆓 Free Tier: Use our public AI node for free out-of-the-box (rate-limited), or switch to your own private API Key for high-performance scanning.

---

🚀 Key Features

1. 🌐 Multi-Chain Monitoring

• Real-time Event Listening: Monitor specific on-chain events including:
  • Transfer (Token movements)
  • PairCreated (Liquidity events on DEXs)
  • OwnershipTransferred (Admin action tracking)
  • ContractDeployment (New contracts)
• Supported Networks:
  • EVM: Ethereum, BNB Chain (BSC), Arbitrum, Base, Polygon.
• Custom RPCs: Configure custom HTTP or WebSocket (WSS) endpoints per network for enterprise-grade latency.
• Auto Source Fetching: Automatically retrieves verified source code via Explorer APIs or acts on raw Bytecode if unverified.

2. 🧠 Specialized Detection Modes

Sentin3l AI offers three distinct audit perspectives:

• 💀 Theft Mode (Attacker View):
  • Simulates an unprivileged external attacker.
  • Focuses on Reentrancy, Overflow, Access Control Bypass, and Logic errors.
  • Ignores centralization risks to reduce noise.
• 🧲 RugPull Mode (Investor View):
  • Simulates a retail investor ("Can I sell?").
  • Detects Honeypots, 100% Taxes, Hidden Mints, Blacklists, and Liquidity Draining capabilities.
• 🛡️ Standard Mode:
  • Comprehensive audit covering both code vulnerabilities and centralization risks (Audit Firm perspective).

3. 🛠️ Powerful Audit Tools

• Live Monitor Stream: A scrolling log showing real-time analysis status of incoming blocks.
• Manual Audit Terminal:
  • Analyze any specific contract address instantly.
  • Code Sandbox: Paste raw Solidity code for static analysis without deployment.
• Interactive AI Chat: Context-aware chat assistant that knows the specific contract's code and audit history.
• AI Contract Generator: Describe your desired logic (e.g., "ERC20 with 2% tax") and get production-ready Solidity code instantly.

4. ⚙️ Advanced Configuration

• Concurrency Control: Set the number of simultaneous AI tasks (1-1000) to manage rate limits.
• Scan Rate Limits: Define how many contracts to analyze per block to save API credits.
• Global Whitelist: Automatically ignores known safe tokens (USDT, WETH, etc.) to focus on unknown risks.
• High Risk Vault: Automatically archives contracts with a safety score ≤ 50 for later review.

---

🛠 Tech Stack

• Frontend: React 19, TypeScript, Vite
• Styling: Tailwind CSS, Lucide React
• Blockchain: Ethers.js v6
• AI Integration: Custom OpenAI-compatible Fetch Client (Streaming Support)
• State Management: React Context + LocalStorage

---

🚦 Getting Started

Prerequisites

• Node.js (v18+)
• Optional: An API Key for an OpenAI-compatible LLM provider. The app works without one (Free Tier), but adding a key improves speed and concurrency.

Installation

1. Clone the repository

   git clone https://github.com/yourusername/sentin3l-ai.git
   cd sentin3l-ai

2. Install dependencies

   npm install

3. Run the development server

   npm run dev

Configuration

1. Free Tier (Default): Just start the app. The "API Key" field in settings will be empty, indicating Free Mode is active.
2. Custom Mode:
   • Open Settings (Gear Icon).
   • Enter your API Key (e.g., OpenAI, DeepSeek, Gemini).
   • Set Base URL (e.g. https://api.openai.com/v1).
   • Increase Concurrency Limit for faster scanning.
   • Click Save Config.

---

📖 Usage Guide

1. Select Networks: Click the "Automatic Monitoring" button. Enable specific networks (e.g., ETH, BSC) and scan types (Transfer, Deployment, Liquidity).
2. Choose Mode: Select Theft, RugPull, or Standard mode from the header.
3. Start Scanning: Click the "Play" button on the network card.
4. Monitor: Watch the stream populate. Click "View" on any contract to see the detailed report.
5. Analyze: Use the Chat to ask specific questions like "Can the owner pause trading?".

---

⚠️ Disclaimer

Sentin3l AI is an experimental security tool.

• Audit results are generated by AI and may contain false positives or negatives ("hallucinations").
• Never rely solely on this tool for financial decisions.
• Always perform manual verification of smart contract code.

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.