Mar 12, 2026
The AI Agency: 112+ Specialized Agents for Any Task
Imagine having an entire agency of AI experts ready to help with any task—frontend development, backend architecture, marketing campaigns, product management, and more. That's exactly what the Agency-agents project offers: a comprehensive collection of 112+ specialized AI agent personas organized into 11 divisions, each designed to integrate seamlessly with popular AI coding tools like Claude Code, Cursor, Aider, and others. Whether you're building a startup MVP or coordinating an enterprise-level project, this open-source repository gives you access to domain experts that understand the nuances of specific fields.
What is Agency-agents?
The Agency-agents project is an open-source GitHub repository created by msitarzewski that provides a complete "AI agency" in a box. The idea came from a Reddit discussion about the need for specialized AI assistants—generic helpers can only take you so far, but true productivity gains come from working with experts who understand the specifics of their domain. The project gained traction quickly, with over 50 requests within the first 12 hours, and has been trending on GitHub ever since.
The repository contains AI agents designed to work alongside your existing tools, not replace them. Each agent has a defined personality, set of rules, workflow process, and success metrics. You don't get a generic assistant that tries to do everything poorly—you get specialists that excel at their particular function.
Key Features
The project delivers on its promise of comprehensive coverage across multiple disciplines:
Comprehensive Agent Library: 112+ specialized agents across 11 divisions covering everything from engineering and design to marketing and spatial computing. The Engineering division alone includes agents like Frontend Wizard, Backend Architect, and DevOps Engineer. Marketing has Growth Hacker, Content Creator, and Twitter Engager. There's even a Specialized division with agents like Reality Checker and Whimsy Injector for those edge cases that don't fit neatly into other categories.
Multi-Tool Integration: Each agent is available in formats compatible with 9+ AI coding tools including Claude Code, GitHub Copilot, Gemini CLI, OpenCode, Cursor, Aider, Windsurf, and OpenClaw. This means you can use these specialized agents regardless of which AI coding assistant you prefer.
Structured Agent Design: Every agent follows a consistent template that includes personality and core mission, critical rules and constraints, technical deliverables and workflow processes, and success metrics and evaluation criteria. This consistency makes it easy to understand what each agent does and how to work with them effectively.
Real-World Workflow Examples: The repository includes complete examples showing how multiple agents collaborate on projects like startup MVP development, marketing campaign launches, enterprise feature development, paid media account optimization, and spatial computing discovery exercises.
How It Works
The Agency-agents project uses a straightforward approach that makes it accessible even if you're not deeply technical:
Markdown-Based Agents: Each agent is defined in a markdown file with structured sections that describe their expertise and working style. You can read through these files to understand exactly what each agent brings to the table before you even install anything.
Conversion System: Bash scripts called convert.sh and install.sh transform the markdown agents into tool-specific formats for your preferred AI coding assistant. This means you don't have to manually reformat anything—the scripts handle the conversion automatically.
Seamless Integration: Once converted, you copy the generated files to your tool's directory and start working with specialized AI experts. The installation script guides you through selecting your tools and generating the appropriate files.
Example: Frontend Wizard Agent
The Frontend Wizard agent specializes in creating beautiful, responsive user interfaces with modern frameworks like React, Vue, and Svelte. Their workflow includes analyzing design requirements and user needs, generating clean and maintainable component code, implementing accessibility best practices, optimizing performance and bundle size, and providing detailed documentation and style guides. This isn't a generalist that knows a little bit about frontend work—it's an agent built specifically for the challenges frontend developers face every day.
Practical Use Cases
The real power of Agency-agents comes from combining multiple agents to tackle complex projects:
Startup MVP Development: Combine Frontend Developer, Backend Architect, Growth Hacker, Rapid Prototyper, and Reality Checker agents to ship your minimum viable product faster. Each specialist handles their domain while the Reality Checker keeps your plans grounded in reality.
Marketing Campaign Launch: Coordinate Content Creator, Twitter Engager, Instagram Curator, Reddit Community Builder, and Analytics Reporter agents for a multi-channel marketing campaign. Instead of one tool trying to handle everything, you have dedicated experts for each platform and function.
Enterprise Feature Development: Use Senior Project Manager, Senior Developer, UI Designer, Experiment Tracker, Evidence Collector, and Reality Checker agents for enterprise-grade delivery. This combination ensures proper documentation, testing, compliance, and stakeholder alignment throughout the development process.
Getting Started
Getting up and running takes just a few steps:
- Clone the repository:
git clone https://github.com/msitarzewski/agency-agents.git - Run the installation script:
cd agency-agents && ./install.sh - Select your tools when prompted—the script generates agent files for your preferred AI coding tools
- Copy the generated files to your tool's agent directory
- Start collaborating with your new AI expert team
Community and Ecosystem
The project has grown beyond its original scope thanks to an active community. A Chinese translation effort has produced 100 translated agents plus 9 China-market originals, maintained by community contributors. The project has also inspired derivative collections like awesome-openclaw-agents. Clear contribution guidelines and active maintenance make it easy for developers to get involved and extend the collection.
Why This Matters
As AI assistants become more integrated into development workflows, specialization matters more than ever. A general-purpose AI helper can assist with many tasks, but it lacks the depth of knowledge that comes from focused expertise. The Agency-agents project provides a ready-to-deploy team of specialists that can augment your capabilities across every aspect of software development and business operations.
If you're a solo developer looking to expand your skills, this collection gives you instant access to expertise you haven't built yet. If you're a startup team needing to punch above your weight, these agents fill gaps without requiring new hires. If you're an enterprise organization seeking to standardize AI-assisted workflows, the structured agent design provides consistency and quality control.
Try it today and experience what it's like to have an entire AI agency working for you.