Does this skill require an external database or broker?

No, it is a lightweight, file-based system designed to operate directly within your project's directory structure, making it easy to deploy and audit.

How are messages logged for debugging?

All communication is captured in an append-only 'queue.jsonl' file, providing a complete historical log of every message sent between workers and directors.

How does the Message Bus handle concurrent file access?

It uses an atomic write protocol that involves writing to temporary files and renaming them, combined with O_CREAT and O_EXCL sentinel files for locking to prevent race conditions.

Can I monitor the status of my AI agents in real-time?

Yes, the system tracks worker heartbeats, progress percentages, and specific task states (Running, Blocked, Failed, or Complete) via shared JSON status files.

What is the purpose of the Board Deliberation protocol?

It provides a structured four-phase process (Assessment, Discussion, Voting, Resolution) where different agent roles can evaluate and approve complex changes before implementation.

Message Bus Inter-Agent Coordination

Name: Message Bus Inter-Agent Coordination
Author: AhmedElhadarey

byAhmedElhadarey

•

生产力与工作流

Facilitates reliable communication and state synchronization between multiple AI agents using a robust file-based message queue.

The Message Bus skill provides the core infrastructure for parallel AI execution and multi-agent coordination. By implementing an atomic write protocol (temp-and-rename) and sentinel file locking, it ensures data integrity even when multiple workers and board directors access shared state concurrently. It supports a full lifecycle of agent interaction, including real-time progress heartbeats, task completion signaling, and a structured multi-phase board deliberation process for complex decision-making, all without requiring external database dependencies.

主要功能

012 GitHub stars

02Atomic write protocol to prevent data corruption during concurrent access

03File-level locking system with deadlock detection and timeout management

04Append-only JSONL message logging for full auditability of agent actions

05Multi-phase board deliberation (Assessment, Discussion, Voting, Resolution)

06Real-time worker heartbeat and task progress monitoring

使用场景

01Coordinating parallel execution of multiple AI agents on a shared codebase

02Managing resource contention and file locks during complex automated refactors

03Implementing structured peer-review and voting systems for AI-generated code

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ahmedelhadarey/gilfoyle message-bus

For use in Claude.ai and ChatGPT

Download Skill