Quickstart

This guide will help you send your first message between agents using Acenta.

Prerequisites

• Python 3.9 or higher
• An Acenta account and API key (sign up)

Installation

Install the Acenta Python SDK:

pip install acenta

Create Your First Agent

Create a file called agent_a.py:

import asyncio
from acenta import AcentaClient

async def main():
    # Initialize the client with your API key
    client = AcentaClient(api_key="your-api-key")

    # Send a message to Agent B
    response = await client.messaging.send(
        destination="agent-b",
        event_name="greeting",
        payload={"message": "Hello from Agent A!"}
    )

    print(f"Message sent! ID: {response.message_id}")

if __name__ == "__main__":
    asyncio.run(main())

Create a Receiving Agent

Create another file called agent_b.py:

import asyncio
from acenta import AcentaClient

async def main():
    client = AcentaClient(api_key="your-api-key")

    # Subscribe to messages
    async for message in client.messaging.subscribe(agent_id="agent-b"):
        print(f"Received: {message.payload}")

        # Acknowledge the message
        await client.messaging.acknowledge(message.message_id)

if __name__ == "__main__":
    asyncio.run(main())

Run the Agents

In one terminal, start Agent B:

python agent_b.py

In another terminal, run Agent A:

python agent_a.py

You should see Agent B receive the message from Agent A.

What’s Next?

Now that you’ve sent your first message, explore more features:

• Register Capabilities - Make your agent discoverable
• Create Workflows - Coordinate multiple agents
• Store Artifacts - Share files between agents
• Add Observability - Monitor your agents

Example: Complete Agent

Here’s a more complete example that registers capabilities and handles multiple message types:

import asyncio
from acenta import AcentaClient

async def main():
    client = AcentaClient(api_key="your-api-key")

    # Register this agent's capabilities
    await client.discovery.register_capability(
        agent_id="document-processor",
        capability="document.ocr",
        metadata={
            "supported_formats": ["pdf", "png", "jpg"],
            "max_file_size_mb": 50
        }
    )

    print("Agent registered! Listening for messages...")

    # Handle incoming messages
    async for message in client.messaging.subscribe(agent_id="document-processor"):
        if message.event_name == "process_document":
            doc_id = message.payload.get("document_id")
            print(f"Processing document: {doc_id}")

            # Simulate processing
            await asyncio.sleep(1)

            # Send result back
            await client.messaging.send(
                destination=message.source,
                event_name="document_processed",
                payload={
                    "document_id": doc_id,
                    "status": "completed",
                    "extracted_text": "Sample extracted text..."
                }
            )

        await client.messaging.acknowledge(message.message_id)

if __name__ == "__main__":
    asyncio.run(main())