Skip to content

Protocols Overview

Communication protocols that enable AI agents to interact with tools, resources, and other agents.

Available Protocols

Model Context Protocol (MCP)

v1.0.0-rc.1 Status: Available

The Model Context Protocol enables AI models to securely access external tools, data sources, and context through a standardized interface.

Key Features:

  • 🔧 Tool Calling - Expose functions for AI models to execute
  • 📦 Resource Access - Provide structured access to data and content
  • 📝 Prompt Templates - Supply templates for generating structured prompts
  • 🔐 Authentication - API Key and OAuth2 support
  • 🚀 Multiple Transports - Stdio and HTTP implementations

→ Go to MCP Documentation

Agent-to-Agent Protocol (A2A)

Planned Status: In Development

The Agent-to-Agent Protocol (formerly ACP) enables communication and collaboration between independent AI agents.

Planned Features:

  • 🤝 Agent Communication - Standardized messaging between agents
  • 📋 Task Management - Task delegation and lifecycle tracking
  • 🔄 Multiple Bindings - JSON-RPC, gRPC, and REST support
  • 🔍 Discovery - Agent discovery and capability negotiation
  • 📊 Streaming - Real-time data streaming between agents

→ View A2A Roadmap

Protocol Comparison

Feature MCP A2A
Purpose Connect AI to tools/resources Agent-to-agent communication
Status ✅ Available 🚧 Planned
Transport Stdio, HTTP JSON-RPC, gRPC, REST
Use Case Tool integration Multi-agent systems
Authentication API Key, OAuth2 TBD

When to Use Which Protocol

Use MCP When:

  • Connecting AI models to external tools and APIs
  • Providing structured data access to language models
  • Building tool providers for AI applications
  • Integrating with Claude Desktop or similar clients
  • Need stdio or HTTP communication

Use A2A When:

  • Building multi-agent systems
  • Enabling agent collaboration and coordination
  • Implementing task delegation between agents
  • Creating agent networks and workflows
  • Need distributed agent architectures

Getting Started

With MCP

[dependencies]
airsprotocols-mcp = { git = "https://github.com/airsstack/airsprotocols" }

See the MCP Quick Start Guide for detailed instructions.

With A2A

Coming soon! A2A is currently in the planning phase.

Protocol Standards

All AIRS protocol implementations follow these principles:

  1. Spec Compliance - Faithful implementation of official specifications
  2. Type Safety - Strong typing and compile-time validation
  3. Async-First - Built on tokio for efficient async operations
  4. Extensibility - Clear extension points for custom implementations
  5. Testing - Comprehensive test coverage

Resources

MCP Resources

A2A Resources

Contributing

We welcome contributions to both protocol implementations! See our Contributing Guide for details.

Current Focus

  • ✅ MCP: Feature complete, accepting enhancements and bug fixes
  • 🚧 A2A: Design and planning phase, architecture discussions welcome

Have questions? Open a discussion or check the documentation for each protocol.