agent-collab

Coordinate multi-agent deliberation through a shared communication board protocol.

1|Updated Jun 16, 2026
One-click install
npx skills add https://github.com/Mattyreed1/fractal-agent-skills --skill agent-collab-mattyreed1
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-collab
Source: https://github.com/Mattyreed1/fractal-agent-skills/tree/main/agent-collab
Command: npx skills add https://github.com/Mattyreed1/fractal-agent-skills --skill agent-collab-mattyreed1

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This skill solves the problem of isolated AI decision-making by providing a structured protocol for multiple agents to debate, challenge, and synthesize information before reaching a final, durable conclusion.

Core Features & Use Cases

  • Shared Collaboration Board: Provides a persistent, multi-party communication layer for agents to post turns, review work, and track thread state.
  • Structured Deliberation: Enforces a four-round protocol (Context, Pushback, Research, Convergence) to ensure blind spots are caught and evidence is verified.
  • Use Case: Use this when you need to reconcile conflicting viewpoints between specialized agents, such as having a researcher and a critic debate a project plan before presenting it to a human.

Quick Start

Start a new collaboration thread named feature-x with participants agent-a and agent-b by posting the initial context.

Frequently Asked Questions about agent-collab

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I facilitate multi-agent deliberation for complex AI decision-making?

Multi-agent deliberation is facilitated through a shared communication board protocol that allows multiple AI agents to debate, challenge, and synthesize information iteratively. This structured approach ensures blind spots are caught before reaching a durable conclusion.

What is the structured protocol for multi-agent collaboration workflows?

The structured collaboration protocol enforces a four-round process: Context, Pushback, Research, and Convergence. This ensures specialized agents review work, provide adversarial debate, verify evidence, and track thread state effectively.

Can I use local JSONL storage for multi-agent collaboration boards?

Yes, the multi-agent collaboration board uses a backend-agnostic architecture compatible with local JSONL storage. This allows persistent, multi-party communication layers to operate without requiring real-time cloud deployments.

When do I need a shared communication board for AI agents?

You need a shared communication board when reconciling conflicting viewpoints between specialized agents, such as having a researcher and a critic debate a project plan. It provides a persistent layer to post turns and track thread state.

Does the multi-agent collaboration protocol work with real-time Convex deployments?

Yes, the multi-agent collaboration protocol implements a backend-agnostic architecture that supports real-time Convex deployments. This allows agents to interact dynamically within the structured four-round deliberation process.

What are the limitations of using a four-round protocol for agent deliberation?

The four-round protocol is limited to structured deliberation workflows requiring iterative feedback and adversarial debate. It is not suited for simple, isolated AI decision-making tasks that do not benefit from multi-party synthesis and evidence verification.