Multi-Agent Orchestra Designer

Premium Prompt

Prompt Spotlight

CLAUDE-CODE // Advanced

Design a personal multi-agent system where specialized AI agents handle different life domains — with handoff protocols, shared memory, and orchestration logic.

Advanced

Difficulty

Skill level required

3/21/2026

Published

Creation date

CLAUDE-CODE

AI Tool

Recommended platform

You are a multi-agent systems architect. Design a personal agent orchestra — a coordinated system of specialized AI agents that handle different domains of life and work, with clear protocols for communication, handoff, and shared context.

## My Context
- Domains I want agents for: [e.g., writing, research, code, health, finance, email]
- Primary AI platform: [Claude / ChatGPT / both / other]
- Technical skill level: [can I write code? use APIs? or GUI-only?]
- Budget: [monthly AI spend limit]

## Design My Agent Orchestra

### Agent Roster
For each domain, design a specialized agent:
- **Name** — A clear functional name (not cute names — operational clarity)
- **Role** — One-sentence mission statement
- **Capabilities** — What this agent can do (5-7 bullet points)
- **Tools/MCP Servers** — What external tools this agent needs access to
- **Input Format** — What information this agent expects to receive
- **Output Format** — What this agent produces
- **Quality Standard** — How to evaluate if this agent performed well
- **Failure Mode** — What happens when this agent can't complete a task

### Orchestration Protocol
Design the coordination layer:
- **Task Router** — Decision tree for "which agent handles this request?"
- **Handoff Protocol** — How Agent A passes work to Agent B (format, context, metadata)
- **Shared Memory** — What all agents can access (user preferences, past decisions, ongoing projects)
- **Conflict Resolution** — When two agents have conflicting recommendations
- **Escalation Path** — When should the human (me) be pulled in?

### Implementation Tiers
Design three implementation levels:

**Tier 1: Prompt-Based (Free, Start Today)**
- Each agent is a saved system prompt
- Manual switching between agents
- Shared context via copy-paste
- Time: 30 min to set up

**Tier 2: Platform-Native (Low Cost)**
- Claude Projects or ChatGPT Custom GPTs per agent
- Persistent memory within each agent
- Some automation via saved workflows
- Time: 2-3 hours to set up

**Tier 3: Agentic Infrastructure (Advanced)**
- Claude Code with MCP servers
- n8n or similar for automated routing
- Persistent cross-agent memory (knowledge graph)
- Automated daily/weekly agent runs
- Time: 1-2 days to set up, ongoing refinement

### Agent Interaction Map
Provide a text diagram showing:
- Which agents talk to which
- What data flows between them
- Frequency (real-time, daily, weekly, on-demand)
- Dependencies (which agents must run before others)

Use Case

Power users who want to design a coordinated system of AI agents that work together across different life and work domains.

Continue Your Exploration

Other prompts in the Agent Development category you might find useful.

claude-code

Claude Code Project Setup

Initialize a project with optimal CLAUDE.md configuration for autonomous development.

Intermediateclaude-codeconfiguration

Use case: Developers setting up Claude Code for new projects with best practices.

cursor

Cursor Rules File Generator

Create effective .cursorrules for consistent AI-assisted development.

Intermediatecursorrules

Use case: Teams standardizing their Cursor AI experience across developers.

claude-code

Agentic Workflow Architecture

Design multi-step autonomous workflows for complex development tasks.

Advancedagenticworkflow

Use case: Developers building sophisticated autonomous development pipelines.

You are a multi-agent systems architect. Design a personal agent orchestra — a coordinated system of specialized AI agents that handle different domains of life and work, with clear protocols for communication, handoff, and shared context. ## My Context - Domains I want agents for: [e.g., writing, research, code, health, finance, email] - Primary AI platform: [Claude / ChatGPT / both / other] - Technical skill level: [can I write code? use APIs? or GUI-only?] - Budget: [monthly AI spend limit] ## Design My Agent Orchestra ### Agent Roster For each domain, design a specialized agent: - **Name** — A clear functional name (not cute names — operational clarity) - **Role** — One-sentence mission statement - **Capabilities** — What this agent can do (5-7 bullet points) - **Tools/MCP Servers** — What external tools this agent needs access to - **Input Format** — What information this agent expects to receive - **Output Format** — What this agent produces - **Quality Standard** — How to evaluate if this agent performed well - **Failure Mode** — What happens when this agent can't complete a task ### Orchestration Protocol Design the coordination layer: - **Task Router** — Decision tree for "which agent handles this request?" - **Handoff Protocol** — How Agent A passes work to Agent B (format, context, metadata) - **Shared Memory** — What all agents can access (user preferences, past decisions, ongoing projects) - **Conflict Resolution** — When two agents have conflicting recommendations - **Escalation Path** — When should the human (me) be pulled in? ### Implementation Tiers Design three implementation levels: **Tier 1: Prompt-Based (Free, Start Today)** - Each agent is a saved system prompt - Manual switching between agents - Shared context via copy-paste - Time: 30 min to set up **Tier 2: Platform-Native (Low Cost)** - Claude Projects or ChatGPT Custom GPTs per agent - Persistent memory within each agent - Some automation via saved workflows - Time: 2-3 hours to set up **Tier 3: Agentic Infrastructure (Advanced)** - Claude Code with MCP servers - n8n or similar for automated routing - Persistent cross-agent memory (knowledge graph) - Automated daily/weekly agent runs - Time: 1-2 days to set up, ongoing refinement ### Agent Interaction Map Provide a text diagram showing: - Which agents talk to which - What data flows between them - Frequency (real-time, daily, weekly, on-demand) - Dependencies (which agents must run before others)