The foundational infrastructure that enables AI agents to communicate, coordinate, and collaborate at scale
π Documentation β’ π Quick Start β’ π― Examples β’ π€ Contributing
English | δΈζ | More Languages Coming Soon...
MPLP (Multi-Agent Protocol Lifecycle Platform) is an enterprise-grade L1-L3 protocol stack that provides the foundational infrastructure for building scalable multi-agent systems. Think of it as the "Internet Protocol Suite" for AI agents - enabling different intelligent agents to communicate, coordinate, and collaborate seamlessly.
MPLP provides the building blocks for multi-agent systems:
| π οΈ MPLP Provides | π€ You Build |
|---|---|
| Standardized communication protocols | Intelligent agents with domain logic |
| Coordination and workflow management | AI decision-making algorithms |
| Resource management and monitoring | Business-specific implementations |
| Security and access control | Industry-specific agent behaviors |
MPLP v1.0 Alpha represents a major milestone in multi-agent protocol development:
- β 100% Feature Complete: All 10 L2 coordination modules implemented
- β Perfect Quality: 2,869/2,869 tests passing (100% pass rate)
- β Enterprise Ready: Zero technical debt, 99.8% performance score
- β Production Tested: Comprehensive security and integration testing
β οΈ API Evolution: APIs may evolve based on community feedback before v1.0 stable
Get up and running with MPLP in under 5 minutes:
# Install from npm
npm install mplp@alpha
# Or clone from source
git clone https://github.com/Coregentis/MPLP-Protocol.git
cd MPLP-Protocol
npm install && npm testimport { MPLPCore, ContextManager, PlanManager } from 'mplp';
// Initialize MPLP protocol stack
const mplp = new MPLPCore({
modules: ['context', 'plan', 'role', 'confirm'],
environment: 'development'
});
// Create a shared context for agent collaboration
const context = await mplp.context.create({
contextId: 'multi-agent-task-001',
participants: ['agent-1', 'agent-2', 'agent-3'],
sharedState: { goal: 'Process customer support tickets' }
});
// Create and execute a collaborative plan
const plan = await mplp.plan.create({
planId: 'support-workflow',
contextId: context.contextId,
goals: [
{ id: 'classify', assignee: 'agent-1' },
{ id: 'route', assignee: 'agent-2' },
{ id: 'respond', assignee: 'agent-3' }
]
});
console.log('Multi-agent collaboration ready! π');- π Read the Complete Documentation
- π― Try the Examples
- ποΈ Build your first Multi-Agent System
MPLP implements a 4-layer protocol stack designed for enterprise-scale multi-agent systems:
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β L4 Agent Layer β
β (Your Intelligent Agent Implementation) β
β π€ AI Decision Logic β’ π§ Learning Algorithms β
β πΌ Business Logic β’ π― Domain-Specific Functions β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β L3 Execution Layer β
β CoreOrchestrator β
β π Workflow Orchestration β’ β‘ Resource Management β
β π System Monitoring β’ π Load Balancing β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β L2 Coordination Layer β
β Context β Plan β Role β Confirm β Trace β Extension β β
β Dialog β Collab β Core β Network β (10 modules total) β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β L1 Protocol Layer β
β π§ Cross-cutting Concerns β’ π JSON Schemas β
β π Security β’ π Performance β’ π State Management β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
| Layer | Purpose | Components | Status |
|---|---|---|---|
| L4 Agent | Your intelligent agents | AI logic, business rules, domain expertise | π― Your implementation |
| L3 Execution | Workflow orchestration | CoreOrchestrator, resource management | β Complete |
| L2 Coordination | Agent coordination | 10 specialized protocol modules | β Complete |
| L1 Protocol | Foundation standards | Schemas, security, cross-cutting concerns | β Complete |
| Module | Purpose | Key Features | Tests |
|---|---|---|---|
| ποΈ Context | Shared state management | Multi-session contexts, state sync | 499/499 β |
| π Plan | Collaborative planning | AI-driven planning, goal decomposition | 170/170 β |
| π€ Role | Access control & RBAC | Enterprise security, permissions | 323/323 β |
| β Confirm | Approval workflows | Multi-party consensus, approval chains | 265/265 β |
| π Trace | Execution monitoring | Performance tracking, audit trails | 212/212 β |
| π Extension | Plugin system | Dynamic extensions, capability expansion | 92/92 β |
| π¬ Dialog | Communication | Inter-agent messaging, conversation flows | 121/121 β |
| π€ Collab | Collaboration | Multi-agent coordination, task distribution | 146/146 β |
| βοΈ Core | Central orchestration | System coordination, resource management | 584/584 β |
| π Network | Distributed communication | Service discovery, network resilience | 190/190 β |
| Quality Metric | Target | Achieved | Status |
|---|---|---|---|
| Test Pass Rate | 100% | 2,869/2,869 | β Perfect |
| Test Suites | All passing | 197/197 | β Perfect |
| Code Coverage | >45% | 47.47% | β Exceeded |
| Performance Score | >95% | 99.8% | β Excellent |
| Technical Debt | Zero | Zero | β Clean |
| Security Tests | All pass | 100% | β Secure |
MPLP enables powerful multi-agent applications across industries:
- Customer Service: Multi-agent support with specialized roles (classification, routing, response)
- Content Moderation: Distributed analysis with human-in-the-loop workflows
- Financial Processing: Multi-stage transaction processing with approval chains
- Supply Chain: Coordinated logistics and inventory management
- Multi-Agent Reinforcement Learning: Coordinated learning environments
- Distributed AI Training: Collaborative model training and optimization
- Agent Swarm Intelligence: Large-scale agent coordination and emergence
- Human-AI Collaboration: Mixed human-agent teams and workflows
- Multi-Agent Simulations: Complex system modeling and simulation
- Distributed Problem Solving: Collaborative optimization and search
- Social Agent Networks: Agent society and interaction research
- Protocol Development: New multi-agent protocol research and testing
Architecture Guide β’ API Reference β’ Protocol Specifications β’ Integration Guide
Getting Started β’ Multi-Agent Patterns β’ Example Applications β’ Best Practices
Contributing Guide β’ Development Setup β’ Testing Guide β’ Release Process
API Stabilization β’ Performance Optimization β’ Enhanced Documentation β’ Production Hardening
Advanced Monitoring β’ Cloud Integration β’ GraphQL Support β’ Mobile SDK
L4 Agent Templates β’ Visual Workflow Designer β’ Advanced AI Integration β’ Enterprise Features
GitHub Repository β’ Documentation β’ Examples β’ Discussions
GitHub Issues for bugs β’ GitHub Discussions for questions β’ Documentation for guides β’ Examples for code samples
We welcome contributions! See our Contributing Guide for code contributions, documentation improvements, bug reports, and community guidelines.
MPLP is released under the MIT License. See the LICENSE file for details.
π Star us on GitHub if MPLP helps your multi-agent projects! π
Built with β€οΈ by the MPLP community