Conduit - Claude, OpenAI (ChatGPT), Gemini API Gateway with Analytics

A transparent, edge-first API gateway for Claude, OpenAI, and Gemini with built-in usage analytics, rate limiting, and beautiful dashboards.

Features

• Transparent Proxy: Forwards Claude, OpenAI, and Gemini API requests without wrapping or modifying the API
• Edge Runtime: Global distribution with Vercel Edge for ultra-low latency
• Rate Limiting: Per-minute and per-day limits enforced at the edge with Vercel KV
• Quota Management: Token-based quotas with real-time tracking
• Usage Analytics: Detailed logging with cost tracking and model breakdowns
• No User Auth: Users view usage by searching with their API key (no login required)
• Admin Dashboard: Full control over API keys, quotas, and analytics
• Streaming Support: SSE streaming responses with real-time token extraction

Quick Start

1. Install Dependencies

npm install

2. Set Up Environment

Create .env.local:

DATABASE_URL=postgresql://...neon.tech/conduit?sslmode=require
KV_REST_API_URL=https://....kv.vercel-storage.com
KV_REST_API_TOKEN=xxx
NEXTAUTH_SECRET=your-secret-32-chars
NEXTAUTH_URL=http://localhost:3000
CRON_SECRET=your-long-random-cron-secret
METRICS_SECRET=your-long-random-metrics-secret
API_KEY_ENCRYPTION_KEY=your-256-bit-hex-key
CLAUDE_API_KEY=sk-ant-your-key-here

3. Set Up Database

npm run db:push

4. Create Admin User

Create your first admin user to access the dashboard:

Interactive mode (recommended):

npm run db:seed

Or specify credentials directly:

npm run db:seed -- --email admin@example.com --password securepass --name "Admin User"

The script will:

• ✅ Validate email format
• ✅ Hash the password securely (bcrypt)
• ✅ Create the admin user in the database
• ✅ Warn if using a weak password

5. Run Development Server

npm run dev

Visit http://localhost:3000

Usage

Admin: Login and Create API Keys

1. Login at http://localhost:3000/login with your admin credentials
2. Go to http://localhost:3000/admin/keys
3. Click "Create New Key"
4. Enter your Claude API key and set limits
5. Copy the generated key (shown once)

Users: View Usage

1. Go to http://localhost:3000/usage
2. Enter your API key
3. View usage stats and remaining quota

Make Claude API Requests

curl -X POST http://localhost:3000/api/claude/v1/messages \
  -H "Authorization: Bearer sk-cond_xxx" \
  -H "Content-Type: application/json" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-3-5-sonnet-20241022",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

### Make OpenAI API Requests (ChatGPT / Responses)

```bash
curl -X POST http://localhost:3000/api/openai/v1/responses \
  -H "Authorization: Bearer sk-cond_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.1-codex-mini",
    "input": "Explain what this repository does."
  }'

Make Gemini API Requests

curl -X POST http://localhost:3000/api/gemini/v1beta/models/gemini-1.5-flash:generateContent \
  -H "Authorization: Bearer sk-cond_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [{ "role": "user", "parts": [{ "text": "Hello!" }] }]
  }'

## Tech Stack

- Next.js 16 + TypeScript
- Neon PostgreSQL
- Vercel KV (Redis)
- Drizzle ORM
- Tailwind CSS

## Project Structure

src/ ├── app/ │ ├── (public)/usage/ # User dashboard │ ├── (admin)/admin/ # Admin dashboard │ ├── api/ │ │ ├── claude/[...path]/ # Main proxy │ │ ├── openai/[...path]/ # OpenAI proxy │ │ ├── gemini/[...path]/ # Gemini proxy │ │ ├── admin/keys/ # Key management │ │ └── usage/ # Usage API │ └── page.tsx # Landing ├── lib/ │ ├── auth/ # API key validation │ ├── proxy/ # Proxy & streaming │ ├── rate-limit/ # Rate limiting │ ├── analytics/ # Usage tracking │ └── db/ # Database └── types/ # TypeScript types

## Deployment

### Vercel (Recommended)

```bash
vercel

Set up Neon PostgreSQL and Vercel KV in Vercel Dashboard > Storage.

Production Checklist

• Set NEXTAUTH_SECRET, API_KEY_ENCRYPTION_KEY, CRON_SECRET, and METRICS_SECRET to strong random values
• Set NEXTAUTH_URL and optionally NEXT_PUBLIC_APP_URL to your canonical HTTPS origin
• Provision Neon PostgreSQL and Vercel KV before enabling traffic
• Run npm run build in CI for every change
• Configure the cron jobs in vercel.json
• Keep .env.local local-only and use .env.example as the shareable template
• Wire a real provider-backed cache warming implementation before enabling that admin workflow

Security Notes

⚠️ For Production:

1. Admin routes are protected with NextAuth and proxy-level gating; restrict admin user creation operationally
2. Rotate encryption keys regularly
3. Never log full API keys
4. Enable monitoring and alerts
5. Use HTTPS everywhere

License

MIT