Skip to content

๐ŸŽญ Explore "The Agency" โ€” 80+ AI Specialist Agent Personas

Status: ๐Ÿ’ก Idea
Priority: ๐Ÿ“Œ Important
Category: ๐ŸŽ“ Learning
Added: 2026-04-01


What

Explore The Agency โ€” a curated collection of 80+ specialised AI agent personas across 8 divisions. Each agent has a unique personality, structured instructions, workflows, deliverables, and success metrics. Study how they're built to inform your own agent design skills.

Why

  • Agent design patterns โ€” learn how to structure agent instructions with personality, scope, tools, and deliverables
  • Directly applicable to Copilot Studio, OpenClaw, and Cloud Cafรฉ persona design
  • 8 divisions cover real-world roles โ€” engineering, design, marketing, product, QA, support, PM, and specialised
  • Open source (MIT) โ€” can adapt patterns for your own agents
  • OpenClaw port exists โ€” same agents already converted to SKILL.md format, so you can test them in OpenClaw too

Key Resources

Resource Link
The Agency (original) github.com/msitarzewski/agency-agents
OpenClaw port (SKILL.md format) github.com/Dev-Dennis-040/openclaw-agency-skills
Works with Claude Code, Cursor, Aider, Windsurf, Gemini CLI, OpenClaw

What's Inside โ€” 80+ Agents Across 8 Divisions

Division Count Highlights
๐Ÿ’ป Engineering 25+ Frontend Dev, Backend Architect, DevOps, Security Engineer, SRE, AI Engineer
๐ŸŽจ Design 8 UI Designer, UX Researcher, Brand Guardian, Whimsy Injector
๐Ÿ“ข Marketing 17 SEO Strategist, Content Creator, Social Media, Reddit Ninja
๐Ÿ“Š Product 4 Product Manager, Growth Strategist
๐ŸŽฌ Project Management 6 Scrum Master, Sprint Planner
๐Ÿงช Testing 8 QA Engineer, Performance Tester, Accessibility
๐Ÿ›Ÿ Support 6 Customer Support, Escalation Handler
๐ŸŽฏ Specialised 15 Smart Contract Dev, Data Engineer, Threat Detection

What to Learn From This

Each agent definition follows a consistent structure worth studying:

๐ŸŽฏ Identity & Personality Traits
๐Ÿ“‹ Core Mission & Workflows  
๐Ÿ› ๏ธ Technical Deliverables (with code examples)
๐Ÿ“Š Success Metrics & Communication Style

This pattern is directly reusable for:

  • โ˜• Cloud Cafรฉ personas โ€” structure barista bot, menu agent, order agent the same way
  • ๐Ÿค– Copilot Studio agents โ€” use the instruction patterns for custom agent instructions
  • ๐Ÿพ OpenClaw agents โ€” use the SKILL.md port to test agents live

Learning Plan (When Ready)

  1. Clone the repo and read 3โ€“4 agent definitions across different divisions
  2. Identify the common structure/pattern used across all agents
  3. Compare the original format vs the OpenClaw SKILL.md port โ€” understand the conversion
  4. Pick one agent (e.g., DevOps Automator) and test it with Claude Code or Copilot CLI
  5. Design a Cloud Cafรฉ agent using the same structured pattern
  6. Content idea: Blog post โ€” "What 80 AI Agent Personas Taught Me About Agent Design"

Connects to other ideas

This pairs perfectly with Explore OpenClaw with ClawPilot Skill โ€” use Agency agents inside OpenClaw for a full hands-on experience.


Added during ClawPilot troubleshooting session โ€” 2026-04-01