📖 Overview

AuraGate is a high-performance, self-hosted AI Gateway Proxy, Smart API Key Rotator, and Token Optimization Engine. Built with Next.js 16, TypeScript, Tailwind CSS, and SQLite (Prisma), AuraGate provides a unified OpenAI-compatible endpoint (/v1) that load-balances requests across multiple LLM providers, automatically handles rate-limiting failovers, reduces token costs, and secures client connections with custom gateway access tokens.

  █████╗ ██╗   ██╗██████╗  █████╗  ██████╗  █████╗ ████████╗███████╗
 ██╔══██╗██║   ██║██╔══██╗██╔══██╗██╔════╝ ██╔══██╗╚══██╔══╝██╔════╝
 ███████║██║   ██║██████╔╝███████║██║  ███╗███████║   ██║   █████╗  
 ██╔══██║██║   ██║██╔══██╗██╔══██║██║   ██║██╔══██║   ██║   ██╔══╝  
 ██║  ██║╚██████╔╝██║  ██║██║  ██║╚██████╔╝██║  ██║   ██║   ███████╗
 ╚═╝  ╚═╝ ╚═════╝ ╚═╝  ╚═╝╚═╝  ╚═╝ ╚═════╝ ╚═╝  ╚═╝   ╚═╝   ╚══════╝

 ⚡ AuraGate v0.1.0 — Smart AI Gateway & API Key Rotator (OpenCode & LLMs)

 > Server Endpoint : http://localhost:20128/v1
 > Web Dashboard   : http://localhost:20128
 > Status          : Ready to route ✓

🎯 Architecture Diagram

flowchart LR
    subgraph Clients["Clients & AI Editors"]
        OC["OpenCode CLI"]
        CU["Cursor IDE"]
        CL["Cline / Roo-Code"]
        CR["cURL / Custom SDK"]
    end

    subgraph AG["AuraGate Engine (Port 20128)"]
        AUTH["Gateway Token Auth (ar-sk-...)"]
        CSP["Centralized System Prompt Injector"]
        RTK["RTK Token Saver & Sanitizer"]
        ROT["Smart Key Pool Rotator & Cooldown"]
        DB[(SQLite DB)]
    end

    subgraph Providers["LLM Providers"]
        GROQ["Groq API"]
        MIST["Mistral AI"]
        OAI["OpenAI"]
        DS["DeepSeek"]
        GEM["Google Gemini"]
    end

    Clients -->|Bearer Token| AUTH
    AUTH --> CSP
    CSP --> RTK
    RTK --> ROT
    ROT <--> DB
    ROT -->|Auto Failover & Round-Robin| Providers

🌟 Key Features

🔄 1. Smart API Key Pool & Automatic Failover

• Round-Robin Key Rotation: Distributes requests across active provider keys sorted by priority.
• Failover Cooldown (429 / 401 / 403 / 5xx): If a key hits a rate limit or authorization error, AuraGate automatically puts it in temporary cooldown (10–30m) and retries the request using the next available candidate key without failing client connections.

✨ 2. Auto-Detect Provider & Live Model Importer

• Provider Auto-Detection: Automatically detects AI providers from key prefixes (gsk_, sk-or-, sk-, AIzaSy, sk-ant-).
• Live Model Syncing: Queries provider /models endpoints to automatically import and register active models.

⚡ 3. RTK Token Saver & Prompt Optimizer

• Whitespace & Linebreak Sanitization: Trims excessive blank lines (\n\n\n+ $\rightarrow$ \n\n) and trailing spaces.
• System Prompt Deduplication: Merges duplicate consecutive system messages.
• In-Memory LRU Caching: Caches identical query responses to save 100% tokens on repeated prompts.

🎯 4. Centralized System Prompt Injection

• Global Rules Engine: Inject global instructions (e.g. "Always respond in Indonesian and write clean TypeScript code") into every chat completion centrally from the Web Dashboard.

🔒 5. Multi-Client Gateway Access Tokens

• Issue secure gateway access tokens (ar-sk-...) for different team members or local editors.
• Protects the proxy endpoint from unauthorized external access.

📦 6. Live Model Catalog & 1-Click Copy

• Search across all imported models from active keys and copy model IDs in 1 click for editor setup.

⚡ 7. Live Provider Network Ping Test

• Test real-time network latency (in ms) to Groq, Mistral AI, OpenAI, DeepSeek, Gemini, and OpenRouter directly from the dashboard.

📊 8. Graph Analytics & Usage Breakdown

• Visual request distribution per provider, average processing latency, and estimated total token savings.

🚀 Quick Start

📦 Option 1: Global NPM Installation (Recommended)

Install AuraGate globally and launch it instantly from any terminal window:

# Install AuraGate globally from NPM
npm install -g auragate

# Launch AuraGate Server & Web Dashboard
auragate

# Or launch on a custom port
auragate -p 8080

Open http://localhost:20128 in your browser to access the Web Admin Dashboard.

🛠️ Option 2: Local Source Installation

# Clone the repository
git clone https://github.com/gugun024/Auragate.git
cd Auragate

# Install dependencies
npm install

# Initialize SQLite database schema
npx prisma db push

# Launch Development Server
npm run dev

🔌 Integration Guides

OpenCode CLI (opencode.json)

{
  "provider": "openai",
  "options": {
    "baseURL": "http://localhost:20128/v1",
    "apiKey": "ar-sk-your-client-gateway-token"
  }
}

cURL

curl -X POST http://localhost:20128/v1/chat/completions \
  -H "Authorization: Bearer ar-sk-your-client-gateway-token" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [{"role": "user", "content": "Halo dari cURL!"}]
  }'

Python SDK (openai)

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:20128/v1",
    api_key="ar-sk-your-client-gateway-token"
)

response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Halo dari Python!"}]
)

print(response.choices[0].message.content)

Node.js SDK (openai)

import OpenAI from 'openai';

const openai = new OpenAI({
  baseURL: 'http://localhost:20128/v1',
  apiKey: 'ar-sk-your-client-gateway-token',
});

async function main() {
  const completion = await openai.chat.completions.create({
    model: 'gpt-4o-mini',
    messages: [{ role: 'user', content: 'Halo dari Node.js!' }],
  });

  console.log(completion.choices[0].message.content);
}

main();

Context7 MCP Integration

Add Context7 MCP server to your editor configuration (opencode.json / claude_desktop_config.json):

{
  "mcpServers": {
    "context7": {
      "command": "npx",
      "args": ["-y", "@context7/mcp-server"]
    }
  }
}

🛡️ Security & Privacy

• Local SQLite Database: Your API keys and client access tokens remain 100% local on your machine in prisma/dev.db.
• .gitignore Protected: .env and SQLite database files are excluded from git commits to prevent accidental token exposure.

📜 License

MIT License © 2026 AuraGate Team