A Coding Guide to Building a Scalable Multi-Agent Communication System Using Agent Communication Protocol (ACP)

This tutorial implements the Agent Communication Protocol (ACP) by building a flexible, ACP-compliant messaging system in Python, utilizing Google’s Gemini API for natural language processing. The guide covers the installation and configuration of the google-generativeai library and introduces core abstractions, message types, performatives, and the ACPMessage data class, which standardizes inter-agent communication.

Core Components of ACP

Message Types

The ACPMessageType enumeration defines the core message categories used in the Agent Communication Protocol, including:

• REQUEST
• RESPONSE
• INFORM
• QUERY
• SUBSCRIBE
• UNSUBSCRIBE
• ERROR
• ACK

Speech Acts

The ACPPerformative enumeration captures the variety of speech acts agents can use when interacting under the ACP framework, such as:

• TELL
• ASK
• REPLY
• REQUEST-ACTION
• AGREE
• REFUSE
• PROPOSE
• ACCEPT
• REJECT

Message Structure

The ACPMessage data class encapsulates all fields required for a structured ACP exchange, including identifiers, participants, performative, payload, and metadata such as protocol version, language, and timestamps. Key methods include:

• to_acp_format: Converts message to standard ACP message format.
• from_acp_format: Parses ACP message from string format.

Agent Implementation

The ACPAgent class represents an autonomous entity capable of sending, receiving, and processing ACP-compliant messages. It manages its own message queue, conversation history, and subscriptions. Key functionalities include:

• Creating messages with create_message
• Sending different types of messages: send_inform, send_query, send_request, and send_reply
• Processing incoming messages with process_message

Message Broker

The ACPMessageBroker serves as the central router for ACP messages, maintaining a registry of agents and a message log. It provides methods to:

• Register agents with register_agent
• Route messages to appropriate recipients with route_message
• Broadcast messages to multiple recipients with broadcast_message

Demonstration of ACP

The demonstrate_acp function orchestrates a hands-on walkthrough of the entire ACP framework by initializing a broker and three distinct agents: Researcher, AI Assistant, and MathBot. It illustrates key interaction scenarios:

• Information Query (ASK performative)
• Action Request (REQUEST-ACTION performative)
• Information Sharing (TELL performative)

Setup Guide

To run the ACP demo in Google Colab:

1. Get your Gemini API Key from here.
2. Replace GEMINI_API_KEY = "YOUR_ACTUAL_API_KEY" in the code.
3. Run demonstrate_acp().

Conclusion

This tutorial implements ACP-based multi-agent systems capable of performing research, computation, and collaboration tasks. The provided sample scenarios illustrate common use cases such as information queries, computational requests, and fact sharing. Readers are encouraged to extend the framework by adding new agent capabilities or incorporating more sophisticated subscription and notification mechanisms.

Download the Notebook on GitHub. All credit for this research goes to the researchers of this project. Also, feel free to follow us on Twitter and join our 95k+ ML SubReddit.