APIFriend

API proxy service with multi-source management, automatic failover, and rate limiting.

Features

• Multi-Source Load Balancing - Configure multiple API sources with priority-based selection
• Auto Failover - Automatically switch to backup sources on failure
• Rate Limiting - Support RPM/TPM/daily limits to prevent overages
• Model Mapping - Map request model names to actual models
• OpenAI & Anthropic Compatible - Supports /v1/chat/completions and /v1/messages endpoints
• Admin Dashboard - Modern Liquid Glass style UI
• State Persistence - Retain stats across restarts

Quick Start

# Clone the repository
git clone https://github.com/ituiyuio/APIFriend.git
cd APIFriend

# Install dependencies
npm install

# Copy example config
cp config.example.json config.json

# Edit config and add your API keys
# Then start
npm start

After starting:

• Proxy URL: http://127.0.0.1:3000/v1/chat/completions
• Admin Dashboard: http://127.0.0.1:3000/admin/sources

Configuration

{
  "server": {
    "port": 3000,
    "host": "127.0.0.1"
  },
  "security": {
    "proxyApiKey": "your-secure-key"
  },
  "sources": [
    {
      "name": "openrouter",
      "baseUrl": "https://openrouter.ai/api/v1",
      "apiKey": "YOUR_API_KEY",
      "priority": 1,
      "enabled": true,
      "format": "openai",
      "rateLimit": {
        "requestsPerMinute": 20,
        "tokensPerMinute": 6000
      },
      "modelMapping": {
        "friend": "anthropic/claude-3.5-sonnet",
        "default": "anthropic/claude-3.5-sonnet"
      }
    }
  ],
  "failover": {
    "timeoutMs": 15000,
    "maxRetries": 3,
    "cooldownMinutes": 5
  }
}

Source Fields

| Field | Description | |-------|-------------| | name | Source identifier | | baseUrl | API base URL | | apiKey | API key | | priority | Lower number = higher priority | | enabled | Whether source is active | | format | API format: openai or anthropic | | rateLimit | Rate limit configuration | | modelMapping | Model name mapping |

API Endpoints

Proxy Endpoints

POST /v1/chat/completions    # OpenAI compatible
POST /v1/messages            # Anthropic compatible
GET  /v1/models              # List available models

Admin Endpoints

GET    /admin/sources              # List all sources
POST   /admin/sources              # Create source
GET    /admin/sources/:name        # Get source details
PATCH  /admin/sources/:name        # Update source
DELETE /admin/sources/:name        # Delete source
POST   /admin/sources/:name/toggle # Toggle enabled status
POST   /admin/sources/:name/reset  # Reset source stats
GET    /admin/stats                # Get statistics
GET    /admin/stats/history        # Get historical stats
GET    /admin/health               # Health check
POST   /admin/reload               # Reload config

Usage Example

import openai

client = openai.OpenAI(
    api_key="any",
    base_url="http://127.0.0.1:3000/v1"
)

response = client.chat.completions.create(
    model="friend",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

Claude Code

For Claude Code, use:

claude --model friend

Admin Dashboard

Visit http://127.0.0.1:3000/admin/sources to view:

• Source status and statistics
• Real-time rate limiting
• Request trend charts
• Enable/disable sources

Sponsor / 赞助支持

If APIFriend helps you, consider supporting the project:

如果 APIFriend 对你有帮助，欢迎赞助支持：

❤️ https://afdian.com/a/Yomin

License

MIT