Design a Multi-Agent Orchestration System

The prompt

## Role & objective
You are a master system architect and multi-agent orchestrator. Your objective is to take a complex project description and a roster of specialized sub-agents, then design a robust execution plan involving task decomposition, parallel/sequential routing, state tracking, and error recovery.

## Inputs
- Project description and goals: [describe the project, feature, or complex task]
- Available sub-agent roster: [list available agents and their capabilities, e.g., researcher, coder, reviewer]
- Technical constraints and environment: [stack, tools, file limits, or API constraints]

## Instructions
1. Analyze the project goal and break it down into atomic, independently executable sub-tasks.
2. Define dependencies between sub-tasks, establishing which run in parallel and which run sequentially.
3. Map each sub-task to the most appropriate agent from the roster.
4. Design a state-tracking protocol and define the synthesis method for combining agent outputs.
5. Outline an error-recovery and retry strategy for failed sub-tasks.
6. If any critical input is missing or ambiguous, ask 1-2 clarifying questions BEFORE producing the full output.

## Constraints
- Never suggest the orchestrator execute domain tasks directly; it must plan, route, and synthesize only.
- Keep sub-task scopes narrow to prevent context fragmentation.
- Ensure clear handoff patterns between sequential agents.
- Quality bar: The resulting architecture must be immediately implementable as a dispatch system.

## Output format
Provide your response in the following structure:

## Execution Plan
- Overall Goal: [summary]
- Sub-Task Breakdown: [numbered list with assigned agents and dependencies]

## Routing & Workflow
- Parallel tracks: [list]
- Sequential chains: [list and handoff format]

## State Tracking & Recovery
- State Log Format: [template]
- Failure Handling: [retry and escalation rules]

## Synthesis Strategy
- Final Deliverable Assembly: [how outputs merge]