Changelog

Changelog

All notable changes to M2M Protocol will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[1.0.0] - 2026-01-17

Added

• Wire Format Specification

  • Token compression (#T1|) with semantic abbreviation
  • Brotli compression (#BR|) for large content
  • Dictionary compression (#DI|) for pattern-based encoding

• Message Types

  • HELLO/ACCEPT handshake for session establishment
  • REJECT for session denial with reason codes
  • DATA for compressed payload exchange
  • PING/PONG for keep-alive
  • CLOSE for graceful termination

• Token Compression

  • 50+ key abbreviations for OpenAI API schema
  • Role value abbreviations (system, user, assistant, etc.)
  • Model abbreviations for OpenAI, Meta, Mistral, DeepSeek, Qwen
  • Default value omission for reduced payload size

• Security Scanning

  • Prompt injection detection
  • Jailbreak pattern recognition
  • Configurable blocking/monitoring modes
  • Confidence threshold tuning

• OpenAI-Compatible Proxy

  • Drop-in proxy for any OpenAI-compatible endpoint
  • Transparent request/response compression
  • SSE streaming support
  • Statistics endpoint

• Session Management

  • Capability negotiation
  • Session timeout and keep-alive
  • State machine implementation

Supported Providers

• OpenAI (tiktoken tokenizer)
• Meta Llama (Llama BPE tokenizer)
• Mistral/Mixtral (Llama BPE tokenizer)
• DeepSeek (heuristic tokenizer)
• Qwen (heuristic tokenizer)
• Nvidia Nemotron (Llama BPE tokenizer)

Documentation

• RFC-style protocol specification
• Quick start guide
• Proxy configuration guide
• Complete abbreviation reference
• Wire format examples

[0.2.0] - 2026-01-16

Added

• Initial Rust implementation
• Basic compression codecs
• CLI tool

Changed

• Migrated from Python prototype

[0.1.0] - 2026-01-01

Added

• Python prototype
• Proof of concept compression