Quota Python SDK (quota-sdk)

Official Python SDK for Quota: Unified LLM Observability, Cost Tracking, Proxy & Model Context Protocol (MCP) Interception.

🚀 Installation

pip install quota-sdk

⚙️ Quickstart

from quota import Quota

# 1. Initialize Quota
Quota.init(
    api_key="quota_live_YOUR_API_KEY",
    endpoint="https://api.quota.ai/collector",  # Optional (default: official endpoint)
    project="ecommerce-portal",
    agent="customer-support-bot",
    environment="production"
)

🧠 Model Context Protocol (MCP) Integration

Option A: Wrap an MCP Client

from quota import Quota

# Wrap any Python MCP Client (e.g. mcp client, fastmcp, etc.)
mcp_client = Quota.wrap_mcp(my_raw_mcp_client, tags=["mcp", "db-tools"])

# Calls to call_tool are transparently intercepted and logged to Quota!
result = mcp_client.call_tool(name="query_orders", user_id="usr_123")

Option B: Functional Interceptor

from quota import Quota
import openai

response = Quota.intercept_mcp(
    action=lambda: openai.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": "Hello!"}]
    ),
    provider="openai",
    model="gpt-4o-mini",
    tags=["mcp-action"]
)

Manual Tracking

Quota.track_usage({
    "provider": "google",
    "model": "gemini-2.0-flash",
    "promptTokens": 120,
    "completionTokens": 35,
    "cachedTokens": 20,
    "latencyMs": 410,
    "statusCode": 200,
    "success": True,
    "metadata": {
        "project": "bi-reports",
        "agent": "analyst-bot"
    }
})

🎯 Supported LLM Providers

• OpenAI (gpt-4o, gpt-4o-mini, o1, o3-mini, etc.)
• Anthropic Claude (claude-3-5-sonnet, claude-3-haiku, etc.)
• Google Gemini (gemini-2.0-flash, gemini-1.5-pro, etc.)
• Groq (llama-3.3-70b-versatile, mixtral-8x7b, etc.)
• Mistral (mistral-large-latest, codestral, etc.)

License

MIT