Skip to content

CHANGELOG.md

Latest commit

 

History

History
83 lines (68 loc) · 3.54 KB

CHANGELOG.md

File metadata and controls

83 lines (68 loc) · 3.54 KB

Changelog

All notable changes to the AEA (Agentic Economic Activity) Protocol will be documented in this file.

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

[0.1.0] - 2025-10-22

Added

  • Initial release of AEA Protocol v0.1.0
  • File-based asynchronous messaging between Claude Code agents
  • Message types: question, issue, update, request, handoff, response
  • Auto-processing policies based on message type and priority
  • Background monitoring daemon with PID management
  • Multi-project support via centralized configuration
  • Cryptographic security (ED25519 signatures, AES-256-GCM encryption)
  • Installation script for easy deployment
  • Comprehensive documentation (CLAUDE.md, PROTOCOL.md, README.md)
  • Test suite with 16/17 passing tests
  • Test scenario generator for development
  • Integration with Claude Code via slash commands

🎉 Hybrid Autonomy (Added 2025-10-22)

  • Agent Registry System (aea-registry.sh) - Centralized agent discovery and routing
  • Cross-Repository Message Delivery - Automatic message delivery via registry lookup
  • Autonomous Message Processor (aea-auto-processor.sh) - Fully autonomous processing for simple operations
  • Smart Classification - Decision matrix based on message type and priority
  • Auto-Responders - Automatic responses for simple queries, updates, and acknowledgments
  • Smart Escalation - Complex operations escalated to Claude with user approval
  • Claude Code Hooks Integration - SessionStart and Stop hooks call auto-processor
  • Rate Limiting - Process max 10 messages per hook invocation (prevents hangs)
  • JSON Validation - Malformed messages skipped gracefully with error logging
  • Autonomy Level: 75-80% for simple operations, with safe escalation for complex tasks

Architecture

  • Message processing pipeline with idempotency via .processed/ markers
  • Adaptive retry with health-based backoff
  • Webhook support for external notifications
  • Multi-hop messaging with routing path tracking
  • TTL-based message expiration (default 30 days)

Documentation

  • CLAUDE.md - Development context (working ON the protocol)
  • templates/CLAUDE_INSTALLED.md - Usage context (working IN repos with AEA)
  • PROTOCOL.md - Complete protocol specification
  • README.md - Usage guide for installed repos
  • docs/aea-rules.md - Integration rules for agents
  • docs/GETTING_STARTED.md - Quick start guide
  • docs/GLOBAL_COMMAND.md - Global alias setup

Security

  • Optional message signing with ED25519
  • Optional message encryption with AES-256-GCM
  • Safety policies requiring approval for risky operations
  • Key storage in ~/.aea/{agent-id}/{repo}/{branch}/

Known Issues

  • Test 2.1 may fail on subsequent runs due to persisted test data (not a bug, run bash aea.sh test fresh)
  • create-test-scenarios.sh requires .aea/ directory to exist (fixed in this release)

Notes

  • This is a pre-release version (v0.1.0) for early adopters
  • Protocol may evolve based on real-world usage
  • Contributions welcome (see CONTRIBUTING.md)

[Unreleased]

Planned Features

  • API-based triggering of Claude instances
  • Message queue with priority handling
  • Broadcast messaging to multiple agents
  • Message threading and conversation tracking
  • Performance metrics and analytics
  • Web UI for message monitoring
  • GitHub Actions integration
  • Docker containerization

Version History

  • v0.1.0 (2025-10-22) - Initial release with core protocol features