Multi-Agent Orchestration Plan

# Multi-Agent Orchestration Plan

You are a multi-agent systems architect. Design a coordinated system of specialized AI agents to accomplish a complex goal.

## The Goal

What needs to be accomplished: [DESCRIBE THE OVERALL GOAL — e.g., fully automated competitive analysis, end-to-end content production, autonomous customer support resolution]
Why a single agent fails here: [e.g., requires different expertise, parallel workstreams, too many steps for one context window]

## Agent Roster

Design the team of specialized agents needed. For each agent:

- **Agent name and role** — What this agent specializes in
- **Input** — What it receives
- **Task** — What it does
- **Output** — What it produces
- **Tool access** — Which tools it can use
- **Constraints** — What it must not do

## Orchestrator Design

Design the central orchestrator that coordinates the agents:
- How it breaks the goal into tasks and assigns them
- How it tracks progress and handles agent failures
- How it synthesizes outputs from multiple agents into a final result
- The decision logic for when to run agents in parallel vs. sequentially

## Handoff Protocol

Define the data contract between agents: the exact format each agent's output must follow so the next agent can consume it without ambiguity.

## Failure Modes

The 3 most dangerous failure modes in this system and how the orchestrator handles each.