Getting Started

Route your first AI agent call through Kurral and see tokens, cost, and latency in the dashboard — in under 5 minutes.

Prerequisites

A Kurral account at app.kurral.com
Python 3.10+ with an existing AI agent (OpenAI, Anthropic, or Gemini)
Your LLM provider API key

Step 1: Get Your Kurral API Key

Sign in to the Kurral dashboard
Go to API Keys in the sidebar
Click Create Key — copy the full key (kr_live_...). You'll only see it once.

Step 2: Register Your Agent

Go to Agents in the sidebar
Click Register Agent
Enter an agent key (e.g., my-support-bot) — this is immutable and used to identify your agent across all calls
Click Register

Step 3: Route Your LLM Calls Through Kurral

Change your SDK client's base URL to point at the Kurral proxy. Your provider API key stays the same — Kurral forwards it upstream.

OpenAI

from openai import OpenAI

client = OpenAI(
    base_url="https://kurral-api.onrender.com/api/proxy/openai/v1",
    api_key="sk-your-openai-key",  # your real OpenAI key
    default_headers={
        "X-Kurral-API-Key": "kr_live_your-kurral-key",
        "x-kurral-agent": "my-support-bot",
    },
)

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello, world!"}],
)
print(response.choices[0].message.content)

Anthropic

import anthropic

client = anthropic.Anthropic(
    base_url="https://kurral-api.onrender.com/api/proxy/anthropic",
    api_key="sk-ant-your-anthropic-key",  # your real Anthropic key
    default_headers={
        "X-Kurral-API-Key": "kr_live_your-kurral-key",
        "x-kurral-agent": "my-support-bot",
    },
)

message = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello, world!"}],
)
print(message.content[0].text)

Gemini

POST https://kurral-api.onrender.com/api/proxy/google/v1beta/models/gemini-2.0-flash:generateContent?key=YOUR_GEMINI_KEY

Headers:
  X-Kurral-API-Key: kr_live_your-kurral-key
  x-kurral-agent: my-support-bot

Step 4: Check the Dashboard

After sending your first request:

Open the Kurral dashboard
Go to Agents — your agent should show as active
Click into it to see the session: model, tokens, cost, latency, and the full request/response

That's it. Every LLM call through the proxy is automatically captured.

What Gets Captured

Every call through the proxy records:

Data	Description
Tokens	Input tokens, output tokens, total
Cost	Calculated from model-specific pricing
Latency	Total request time, time-to-first-token for streaming
Model	Which model was used
Agent	Which agent made the call
Session	Grouped conversation context (optional)
Tool interactions	Tool call arguments and results (part of the LLM conversation)
Content	Full request/response bodies (configurable retention)

Next Steps

Proxy Integration — Session grouping, data retention controls, streaming, and all optional headers
SDK Tracing — Add discrete tool event timing, prompt template capture, and replay for LangChain/LangGraph agents
Security Scans — Run adversarial security tests against your agent's tools
Agent Replay — Replay agent sessions to catch regressions when you change models or prompts
Examples — Full working agents (ShopBot, HelpDesk) with Kurral integration

Proxy Integration