What happens when two agents disagree on a design choice?

The protocol triggers a two-round debate where each agent states their rationale. If a compromise isn't reached after two rounds, the Orchestrator escalates the decision to the user.

What are workflow gates?

Gates are execution checkpoints that require an agent to verify specific prerequisites (like a passing test or a domain review) before proceeding to the next phase of development.

How does the 'Zero Memory' principle affect agent performance?

It ensures reliability. Since agents do not share a global history, the protocol mandates providing absolute paths, exact error messages, and specific requirements in every delegation to prevent context loss.

What is the primary role of the Orchestrator in this protocol?

The Orchestrator acts as a project manager that decides what needs to be done and which specialized agent should do it; it never performs specialized work like writing code or editing files directly.

Multi-Agent Orchestration Protocol

Name: Multi-Agent Orchestration Protocol
Author: jwilger

byjwilger

•

Productivity & Workflow

Orchestrates specialized agents through structured delegation, workflow gates, and formal conflict resolution patterns.

The Orchestration Protocol provides a robust framework for managing complex multi-agent workflows within Claude Code. It establishes clear boundaries where a central orchestrator delegates specialized tasks to focused agents—such as TestWriters, Implementers, or Domain Modelers—ensuring that no single agent becomes a logic bottleneck. By implementing mandatory workflow gates, 'zero-memory' context templates, and formal conflict resolution protocols including domain expert vetos, this skill maintains high-integrity code development and prevents autonomous agent drift or architectural inconsistencies.

Key Features

01Specialized Agent Delegation: Separates coordination logic from domain-specific implementation tasks.

02Mandatory Workflow Gates: Enforces discipline by preventing the skipping of critical steps like testing or review.

037 GitHub stars

04Zero-Memory Context Injection: Standardized templates to ensure agents receive complete task data every time.

05State Tracking: Manages lifecycle transitions between testing, type definition, and implementation phases.

06Formal Conflict Resolution: A structured 2-round debate protocol for resolving agent disagreements.

Use Cases

01Resolving design pattern disputes between agents using structured escalation and domain veto power.

02Managing a full TDD cycle across specialized TestWriter, DomainModeler, and Implementer agents.

03Coordinating complex architectural changes that require synchronized updates across multiple layers.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jwilger/claude-code-plugins orchestration-protocol

For use in Claude.ai and ChatGPT

Download Skill