EMCL v0.1 – Secure Protocol Layer for AI Tool Communication

This is the initial public release of EMCL (Encrypted Model Context Layer) — a lightweight, secure protocol designed to extend Model Context Protocol (MCP) with production-ready security primitives.

EMCL is to MCP what TLS is to HTTP — it adds encryption, signing, and identity protection for AI agent–tool interactions using JSON-RPC-like messaging.

🚀 What's Included
🔐 AES-256-GCM encryption of tool inputs/outputs

✅ HMAC-SHA256 signing of request metadata

🔑 JWT-based agent identity propagation

🛡 Replay protection via timestamp and nonce

📜 Policy enforcement and audit log recommendations

🌐 Gateway + example client + mock tool

📄 Full protocol spec in spec/EMCL-v0.1.md

🧪 Working example: example_client.py → gateway → tool_server

📦 How to Use
Run the included gateway/server.py (EMCL Gateway)

Start the mock_tool_server.py to simulate a backend tool

Run example_client.py to send a full EMCL-compliant encrypted request

📚 Learn More
See the specification and README.md for full details.

📘 License
MIT

👤 Author
Created by Balachandar Manikandan, EMCL Protocol Maintainer
GitHub: @Balchandar

Full Changelog: https://github.com/Balchandar/emcl-protocol/commits/v0.1.0