<\!-- Hero Section -->

The Agent Mesh Communication Protocol

Build scalable, intelligent multi-agent systems with agent mobility, LLM orchestration, and event-driven architecture. Deploy on Kubernetes, Quarkus, Kafka, and NATS. Open-source framework for startups and enterprises.

Watch Demo Explore Concepts GitHub
<\!-- Featured Video Section --> <\!-- Core Concepts Section -->

Three Core Agent Types

💬

Chat Agent

LLM-powered conversational agents. Handle user interactions, process natural language, and orchestrate complex dialogues across the mesh.

  • ✓ Multi-turn conversations with state
  • ✓ Context preservation & memory
  • ✓ Tool integration (MCP/A2A)
  • ✓ Multi-model LLM support
  • ✓ Real-time streaming responses
🎼

Orchestrator Agent

Coordinates complex workflows across multiple agents. Manages task distribution and state synchronization.

  • ✓ Workflow orchestration
  • ✓ Task distribution
  • ✓ Error recovery
🕸️

Mesh Agent

Infrastructure agents managing the mesh. Handle routing, discovery, monitoring, and reliable communication.

  • ✓ Event routing
  • ✓ Agent discovery
  • ✓ Load balancing
<\!-- Agent Mobility Section -->

🚀 Agent Mobility

AMCP implements IBM Aglet-style strong mobility – agents migrate across contexts at runtime with full state preservation. Deploy in the cloud, move to the edge, all state intact.

Key Capabilities:

  • ✓ State serialization & restoration
  • ✓ Subscription migration
  • ✓ Zero message loss guarantee
  • ✓ Agent cloning for redundancy

Migration Flow

1. Save State
onDeactivate() → serialize
2. Transfer
Network → Context B
3. Activate
onLoad() → resume
<\!-- LLM Orchestration Section -->

🧠 LLM Orchestration & Multi-Agent Chat

📊 Multi-Model Support

Integrate OpenAI, Anthropic, LLaMA. Route requests by complexity, cost, or latency.

🔗 Tool Integration (MCP)

Connect to external tools via Model Context Protocol with secure auth.

💭 Reasoning & Planning

Agents collaborate using chain-of-thought reasoning and multi-step planning.

🧠 Context Management

Maintain conversation context across agents with efficient memory management.

⚡ Real-time Streaming

Stream LLM responses in real-time for responsive user experiences.

🛡️ Safety & Guardrails

Guardian agents enforce policies and audit all LLM interactions.

<\!-- Deployment Technologies Section -->

☁️ Scale with Modern Infrastructure

☸️

Kubernetes

Native K8s with Helm charts and auto-scaling.

⚡

Quarkus

Sub-second startup, minimal memory.

📨

Kafka

High-throughput event streaming.

🚀

NATS

Ultra-fast messaging for edge.

🏗️

Microservices

Independently deployable agents.

🔐

Service Mesh

Istio & Linkerd with mTLS.

<\!-- v1.6 Features Section -->

🎯 AMCP v1.6 Major Features

📊 Advanced Monitoring

Real-time insights with event throughput, latency heatmaps, and error rates.

  • ✓ Real-time agent status
  • ✓ Distributed tracing
  • ✓ Performance heatmaps

🎮 Management API

Full programmatic control with REST/gRPC APIs.

  • ✓ Agent lifecycle control
  • ✓ Policy management
  • ✓ Event injection

🔐 Enhanced Security

Enterprise-grade security with audit trails.

  • ✓ Structured audit logs
  • ✓ Policy enforcement
  • ✓ Compliance reporting

🌍 Multi-Tenancy

Secure isolation for SaaS deployments.

  • ✓ Tenant isolation
  • ✓ Usage tracking
  • ✓ Resource quotas
<\!-- Technical Specifications Section -->

Technical Specifications

📋 Protocol Specification

Event-driven async pub/sub with guaranteed delivery semantics. Supports exactly-once and at-least-once delivery modes.

🔐 Security Model

mTLS encryption, RBAC, OAuth2 integration, audit logging, and compliance reporting for enterprise deployments.

📊 Observability

Distributed tracing, metrics collection, health checks, and real-time monitoring dashboards.

🔄 Agent Mobility

IBM Aglet-style strong mobility with state serialization, subscription migration, and zero message loss.

🌐 Multi-Broker

Support for Kafka, NATS, RabbitMQ, and custom brokers with pluggable transport layer.

⚙️ Extensibility

Plugin architecture for custom agents, brokers, serializers, and middleware components.

<\!-- Open Source Contribution Section -->

🚀 Join the Open-Source Community

AMCP is an open-source project. We welcome contributions from developers, researchers, and enterprises building the future of agent-based systems.

How to Contribute

  • 📝 Documentation
    Help improve guides, API docs, and tutorials for the community.
  • 💻 Code
    Submit pull requests for features, bug fixes, and performance improvements.
  • 🧪 Testing
    Write tests, report issues, and help ensure quality and reliability.
  • 🎓 Examples
    Create sample applications and use case demonstrations.
<\!-- CTA Section -->

Ready to Build Your Agent Mesh?

Start with AMCP v1.6 today. Open-source, enterprise-ready, and built for scale. Join thousands of developers building intelligent agent systems.

Get Started View on GitHub Watch Demo