Quick Start

pip install dakora-client

export DAKORA_BASE_URL="https://api.dakora.io"    # or your deployment URL
export DAKORA_API_KEY="dkr_..."                   # your project API key
export OPENAI_API_KEY="sk-..."                    # your OpenAI key (for agent examples)
export OPENAI_CHAT_MODEL_ID="gpt-4o-mini"                 # model for OpenAI agent client

$env:DAKORA_BASE_URL = "https://api.dakora.io"
$env:DAKORA_API_KEY = "dkr_..."
$env:OPENAI_API_KEY = "sk-..."
$env:OPENAI_CHAT_MODEL_ID = "gpt-4o-mini"

import asyncio
from dotenv import load_dotenv
from dakora_client import Dakora

# Load environment variables from .env file
load_dotenv()


async def main():
    # uses env vars: DAKORA_API_KEY and DAKORA_BASE_URL
    # initialize Dakora client
    dakora = Dakora()

    # Render a Dakora template and run it through the agent (auto-tracked)
    faq_template = await dakora.prompts.render(
        "faq_responder",
        {
            "question": "What is Dakora?",
            "knowledge_base": "**Dakora** is an open‑source AI observability and prompt management platform that delivers real‑time cost analytics, policy controls, and a hosted studio to help teams make every token count. Learn more at https://dakora.io.",
            "tone": "helpful",
            "include_sources": True,
        },
    )

    print("Rendered template:\n\n", faq_template.text)


asyncio.run(main())

import asyncio
from dotenv import load_dotenv
from agent_framework import ChatAgent
from dakora_client import Dakora
from dakora_agents.maf import DakoraIntegration
from agent_framework.openai import OpenAIChatClient

# Load environment variables from .env file
load_dotenv()

async def main():
    # uses env vars: DAKORA_API_KEY and DAKORA_BASE_URL
    # initialize Dakora client
    dakora = Dakora()

    # One-line observability setup (OTel via OTLP/HTTP → Dakora)
    middleware = DakoraIntegration.setup(dakora)

    # uses env vars: OPENAI_API_KEY and OPENAI_CHAT_MODEL_ID
    # Create your agent with middleware attached
    agent = ChatAgent(
        id="dakora_faq_agent",
        name="DakoraFAQAgent",
        chat_client=OpenAIChatClient(),
        middleware=[middleware],
    )

    # Render a Dakora template and run it through the agent (auto-tracked)
    faq_template = await dakora.prompts.render(
        "faq_responder",
        {
            "question": "What is Dakora?",
            "knowledge_base": "**Dakora** is an open‑source AI observability and prompt management platform that delivers real‑time cost analytics, policy controls, and a hosted studio to help teams make every token count. Learn more at https://dakora.io.",
            "tone": "helpful",
            "include_sources": True,
        },
    )

    # Run the agent with the rendered template
    reply = await agent.run(faq_template.text)
    print(reply)

    # Optional: ensure spans are flushed before exit
    DakoraIntegration.force_flush()

asyncio.run(main())