Synapse A2A (Agent-to-Agent) is a framework that allows different AI models and instances to communicate, share tasks, and collaborate on complex projects within the Claude Code environment.

How do I define delegation rules?

Rules are defined in the .synapse/delegate.md file using YAML frontmatter. You can specify match conditions like keywords or file patterns to determine which agent receives a specific task.

Can I monitor the status of my delegated tasks?

Yes, the 'synapse list' command provides a real-time interactive TUI that displays whether agents are READY, PROCESSING, or WAITING, allowing you to jump to specific sessions or kill processes.

What is the difference between Roundtrip and Oneway flow?

Roundtrip mode waits for the delegated agent to complete the task and return a response to the sender, whereas Oneway is a 'fire and forget' mode used for background tasks.

Is it safe for multiple agents to edit the same files?

The Delegation skill integrates with File Safety locks to prevent conflicts. It verifies file locks before delegating tasks to ensure that only one agent modifies a file at a time.

Synapse Agent Delegation

Name: Synapse Agent Delegation
Author: s-hiraoku

bys-hiraoku

コラボレーションツール

Automates task routing and coordination between specialized AI agents using customizable natural language rules and priority levels.

概要

The Delegation skill enables Claude to act as an orchestrator or a participant in a multi-agent ecosystem via Synapse A2A. It allows users to define sophisticated routing rules based on keywords and file patterns, sending specific tasks like research to Gemini, coding to Codex, or reviews to other Claude instances. With support for both 'orchestrator' and 'passthrough' modes, interactive status monitoring, and integrated file safety locks, this skill transforms Claude from a solo assistant into a powerful coordinator of specialized AI workers.

主な機能

Support for Orchestrator, Passthrough, and Manual delegation modes
0 GitHub stars
Priority-based task queuing and agent naming for custom roles
Real-time agent status monitoring with an interactive TUI
Rule-based task routing via YAML and Markdown configuration
Seamless integration with File Safety locks and Git branch verification

ユースケース

Orchestrating complex dev workflows where research, implementation, and review are handled by different specialized agents
Managing a fleet of AI agents with specific priorities to handle urgent bug fixes vs. background documentation updates
Automating repetitive coding tasks by routing them to a 'Codex' agent while the user continues high-level planning

概要

主な機能

Support for Orchestrator, Passthrough, and Manual delegation modes
0 GitHub stars
Priority-based task queuing and agent naming for custom roles
Real-time agent status monitoring with an interactive TUI
Rule-based task routing via YAML and Markdown configuration
Seamless integration with File Safety locks and Git branch verification

ユースケース

Orchestrating complex dev workflows where research, implementation, and review are handled by different specialized agents
Managing a fleet of AI agents with specific priorities to handle urgent bug fixes vs. background documentation updates
Automating repetitive coding tasks by routing them to a 'Codex' agent while the user continues high-level planning