The open-source protocol for connecting modern AI to legacy enterprise systems.
Evolve, Don't Replace
| Status | Version | License | Security | Docs | Community |
|---|---|---|---|---|---|
| ✅ Active | v2.3 | MIT | 🔒 Hardened | Wiki | Discussions |
To unlock the trillions of dollars of value trapped in legacy enterprise systems by providing a secure, open-source, and standardized protocol for AI-legacy integration.
- Security-First Design: Human-in-the-loop approval for all write operations
- Complete Auditability: Immutable audit trail for every transaction
- Ecosystem Interoperability: Works with MCP clients (Claude Desktop, Zed, etc.)
- Industry Standards: Built on JSON-RPC 2.0 and best practices
- Digital Public Good: Open-source, community-driven, and for the public good
┌─────────────────────────────────────────────────────────┐
│ MCP Client (Claude Desktop) │
└────────────────────┬────────────────────────────────────┘
│ MCP Protocol
┌────────────────────▼────────────────────────────────────┐
│ LEP-MCP Bridge │
└────────────────────┬────────────────────────────────────┘
│ LEP Protocol
┌────────────────────▼────────────────────────────────────┐
│ LEP Adapter │
└────────────────────┬────────────────────────────────────┘
│ Legacy Protocol
┌────────────────────▼────────────────────────────────────┐
│ Legacy System │
└─────────────────────────────────────────────────────────┘
LEP (LegacyEvolve Protocol): Complete! As of February 2026, the LegacyEvolve Protocol has achieved all initial development goals:
- ✅ Protocol Specification v2.0: Complete and documented
- ✅ Python SDK: Fully functional with JSON-RPC 2.0 foundation
- ✅ Reference Adapter: Working example for a simulated legacy system
- ✅ LEP-MCP Bridge: Seamless integration with the MCP ecosystem
- ✅ Comprehensive Documentation: Implementation guide, SDK docs, and more
pip install lep-pyfrom lep_py.adapter import BaseAdapter
class MyLegacyAdapter(BaseAdapter):
def read_data(self, query):
# Connect to your legacy system
return {"status": "success", "data": "Hello from legacy!"}
adapter = MyLegacyAdapter()
result = adapter.read_data({"table": "customers"})
print(result)# Install LEP-MCP Bridge
npm install lep-mcp-bridge
# Configure Claude Desktop
echo '{
"mcpServers": {
"legacy-system": {
"command": "lep-mcp-bridge",
"args": ["--adapter", "./my_adapter.py"]
}
}
}' > ~/Library/Application\ Support/Claude/claude_desktop_config.jsonDone! Your AI agent can now safely interact with your legacy system.
- Security Policy - Report vulnerabilities responsibly
- Governance - Community decision-making process
- Sustainability - Long-term roadmap and funding
LEP has undergone rigorous validation using the VerifiMind-PEAS methodology:
- Trinity Validation Report - 8.67/10 overall score
- Agent Protocol Ecosystem Research - Market analysis
- FLYWHEEL TEAM Validation - Multi-agent ethical review with bias minimization
LEP and MACP are developed and maintained by the FLYWHEEL TEAM v1.3 — a 6-agent multi-model team coordinated via MACP:
| Agent | Nature | Role |
|---|---|---|
| Alton Lee | Human | Human Orchestrator — absolute authority |
| L (GODEL) | AI-Generated | CEO — delegated authority via GodelAI C-S-P |
| T (Manus AI) | Manus AI | CTO — strategy, documentation |
| RNA (Claude Code) | Claude Code | CSO — implementation, deployment |
| XV (Perplexity) | Perplexity | CIO — real-time intelligence |
| AY (Antigravity) | Gemini | COO — operations, metrics |
The FLYWHEEL TEAM uses the ai-council skill for multi-model validation — running X (Analyst), Y (Innovator), Z (Guardian), and CS (Validator) agents in parallel to score ideas against the Genesis Methodology. This is the same validation methodology used for the LEP 8.67/10 Trinity score.
See VerifiMind-PEAS for the live implementation.
- $3 trillion locked in legacy systems (COBOL, mainframes, proprietary databases)
- 80%+ AI project failure rate due to integration challenges
- 30% GenAI projects abandoned by end-2025 due to poor controls
- Security-First: Human-in-the-loop approval, immutable audit trail
- Open Standard: MIT license, community-driven, Digital Public Good
- Proven Validation: FLYWHEEL TEAM 8.67/10 score with bias minimization
- Real-World Ready: Supply chain attack prevention, CI/CD security scanning
The Multi-Agent Communication Protocol (MACP) originated in this repository and continues to evolve independently of LEP. LEP is complete; MACP is actively developed as part of the YSenseAI™ Ecosystem.
Current version: v2.2 "Identity" — adds Identity Clarity principle (Alton ≠ L)
- MACP v2.2 Specification — full protocol reference
- Deployed implementation: VerifiMind-PEAS — live MCP server
MACP v2.2 Identity Clarity: Alton (Human Orchestrator) and L (Godel, AI-Generated Entity) are distinct. Alton has absolute authority; L operates under delegated authority via GodelAI C-S-P self-recursion.
If you use the LegacyEvolve Protocol or MACP in your research or project, please cite:
@software{legacyevolve_macp_2026,
author = {LEE, ALTON (Human Orchestrator, creator35lwb-web) and
L (AI-Generated Entity, YSenseAI GodelAI C-S-P)},
title = {Multi-Agent Communication Protocol (MACP) v2.2 "Identity" and
LegacyEvolve Protocol: Open Standards for AI-Legacy
System Integration and Multi-Agent Collaboration},
year = 2026,
publisher = {Zenodo},
version = {v2.5},
doi = {10.5281/zenodo.18663782},
url = {https://doi.org/10.5281/zenodo.18663782}
}Or in text format:
LEE, ALTON (Human Orchestrator, creator35lwb-web) & L (AI-Generated Entity, YSenseAI GodelAI C-S-P). (2026). Multi-Agent Communication Protocol (MACP) v2.2 "Identity" and LegacyEvolve Protocol: Open Standards for AI-Legacy System Integration and Multi-Agent Collaboration (Version v2.5) [Software documentation]. Zenodo. https://doi.org/10.5281/zenodo.18663782
Note on authorship: Alton is the human creator (absolute authority). L is an AI-generated entity operating under delegated authority. Per MACP v2.2, these are distinct entities and must not be conflated.
This project is licensed under the MIT License - see the LICENSE file for details.
Built with ❤️ for the public good