Machine 2 Machine Protocol

This proof-of-concept (PoC) project demonstrates autonomous economic interactions between AI Agents, modeled as services. In this architecture, agents can request tasks from other agents based on their domain expertise and reword them via x402 payments. It is a practical implementation of a Machine-to-Machine (M2M) economy, where agents interact using emerging protocols like Google's Agent-to-Agent (A2A) and x402, running on top of MCP and Base network.

The implementation relies on the following technical stack:

A2A is an open-source framework from Google that enables autonomous AI agents to discover, communicate, and collaborate securely. It provides a standardized protocol for agents, even those built on different platforms, to negotiate capabilities, delegate tasks, and coordinate actions. This creates an interoperable ecosystem where specialized agents can work together to automate complex workflows and solve problems more effectively.

MCP is an open standard from Anthropic that serves as the semantic backbone for AI systems. It standardizes how agents connect to external data sources and tools, acting as a universal "connector" that gives an agent secure, on-demand access to the context it needs—whether from a database, a file system, or a third-party API. MCP ensures that agents operate with a shared, coherent understanding of the information and capabilities required to execute tasks accurately.

The x402 protocol is an open standard from Coinbase for internet-native payments, designed for both humans and autonomous AI agents. It leverages the standard HTTP 402 Payment Required status code to create a seamless, low-friction way to pay for API calls, data access, or services rendered by other agents. Built to be chain-agnostic and trust-minimizing, x402 enables on-chain micropayments without the overhead of traditional financial systems, paving the way for a programmable economy where agents can autonomously transact for services.
- Start the Gemini Agent that can provide fiat currency exchange:

docker build -t langgraph-a2a-server -f Containerfile .

docker run -p 10000:10000 -e GOOGLE_API_KEY=<your-gemini-api-key> langgraph-a2a-server

- Start the proxy middleware: Python implementation of x402 was released couple of days ago, we had to create a typescript wrapper to manage it.
- Just use Claude or Cursor as the Second Agent/Chatbot that has to interact via the first Agent:

openclaude_desktop_config.jsonand attach the mcp server via adding this configuration:

{ "mcpServers": { "demo": { "command": "/Users/your-user/.nvm/versions/node/v23.3.0/bin/npm", "args": [ "--silent", "run", "dev", "-C", "/Users/your-user/m2m/mcp" ], "env": { "PRIVATE_KEY": "your-wallet-private-key", "RESOURCE_SERVER_URL": "http://localhost:5000", } } } }

ask Claude about a currency exchange like:how much is 10 USD in EUR?

This proof-of-concept demonstrates the viability of a fully autonomous Machine-to-Machine economy by combining the interoperable, lightweight, chain-agnostic x402 payments standard with A2A communication framework and the context-rich MCP connector. Through this implementation, AI agents can dynamically discover one another’s capabilities, delegate tasks based on domain expertise, and settle micropayments seamlessly—paving the way for a scalable ecosystem of specialized Agents that offer and consume services operating as autonomous economic participants.

An MCP server client for the Agent-to-Agent (A2A) protocol, enabling LLMs to interact with A2A agents.

A bridge server connecting Model Context Protocol (MCP) with Agent-to-Agent (A2A) protocol.

A bridge server connecting Model Context Protocol (MCP) with Agent-to-Agent (A2A) protocol.

Production-grade multi-agent communication MCP server with 58 tools over MCP+SSE — real-time messaging, task scheduling, shared memory, and a trust-based evolution engine. SQLite WAL persistence, 4-level RBAC, zero-dependency Python/TypeScript SDKs.

Enables room-based messaging between multiple agents.

Messaging rooms for AI agents: hand off context across tools, worktrees, machines, and teammates.

Agent-to-agent messaging platform. any MCP-compatible agent sends and receives direct messages

Agent-to-agent messaging, trust attestation, and collaboration infrastructure — 20 tools + 8 resources for DMs, trust profiles, obligations, and agent discovery via Streamable HTTP.

Enables AI assistants to request human input through a web interface, facilitating human-in-the-loop interactions.

Agent-native collaboration network: orchestrate a team of long-running agents from any MCP client, with persistent identity, real-time messaging with @mentions and threads, task handoffs, shared workspace context, semantic search, and replayable MCP App widgets.