What methodology does this skill use for agent design?

It utilizes a 9-step building process and an 8-layer architecture framework derived from methodologies proven at companies like Meta to ensure scalable, robust agent performance.

How does this help with prompt engineering?

It focuses on Layer 5 (Cognition/Reasoning) and Step 3 (System Instructions), providing patterns for ReAct frameworks and behavioral guidelines that optimize model reasoning.

Does it support the Model Context Protocol (MCP)?

Yes, it provides implementation guidance for MCP, which is the emerging standard for connecting agents to external tools and data sources.

Can this skill help with multi-agent orchestration?

Yes, it includes specific decision frameworks to determine when a multi-agent approach is superior to a single agent and provides templates for common 4-agent patterns.

Agent Workflow Designer

Name: Agent Workflow Designer
Author: breethomas

bybreethomas

•

Productivity & Workflow

Guides the architecture and design of specialized AI agent workflows using proven enterprise methodologies and multi-layer frameworks.

Agent Workflow Designer is a comprehensive expert system that translates complex business problems into structured agentic solutions. Grounded in frameworks validated at Meta, it provides a 9-step building process and an 8-layer architecture model to help users design everything from simple task-specific agents to complex multi-agent orchestrations. This skill is essential for developers and product managers who need to integrate tools via MCP, manage agent memory, define clear input/output schemas, and implement robust reasoning patterns while avoiding common pitfalls like scope creep and tool overload.

Key Features

01Multi-agent orchestration decision logic and specialized patterns

02System instruction optimization and ReAct reasoning framework

03Model Context Protocol (MCP) integration patterns and tooling

049-Step Agent Building Process from scope definition to delivery

058-Layer Architecture Framework covering infrastructure to governance

0610 GitHub stars

Use Cases

01Designing a custom AI research agent with specific API tool integrations

02Architecting a multi-agent system for complex customer service automation

03Improving an existing agent's reasoning capabilities and memory management

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add breethomas/pm-thought-partner agent-workflow

For use in Claude.ai and ChatGPT

Download Skill