The Agency is an open-source AI agent framework released on GitHub. It organizes 232 specialized AI agents across 16 divisions, each equipped with unique personality, workflows, and deliverable definitions. Licensed under MIT, it supports one-command installation into tools such as Claude Code, Cursor, and Gemini CLI.

📑Table of Contents
  1. What is The Agency
  2. Supported Tools and Installation Steps
  3. Structure of 16 Divisions and 232 Agents
  4. Real-World Use Cases and Examples
  5. Considerations and Limitations When Adopting
  6. Frequently Asked Questions (FAQ)
  7. Comparison Table: Supported Tools
  8. Summary

What is The Agency

The Agency (Agency Agents) is a collection of AI agents developed by msitarzewski. It aims to scale a single AI assistant into an “organization” level system. Each agent includes elements like Identity & Memory, Core Mission, Critical Rules, Technical Deliverables, Workflow Process, and Success Metrics.

The official GitHub repository contains over 10,000 lines of personality, process, and code examples. It gained 50K stars within two weeks of the initial release, indicating strong community interest. Full details are available at the official repository (https://github.com/msitarzewski/agency-agents).


Supported Tools and Installation Steps

The Agency supports multiple AI coding tools. Installation is simplified through provided scripts.

Clone the repository and run install.sh with a tool flag, such as . For Cursor, agents are converted to .mdc files and placed in .cursor/rules/. Similar conversion scripts exist for Gemini CLI and OpenCode, supporting parallel execution.

After installation, agents are ready to use once placed in the appropriate tool directories.


Structure of 16 Divisions and 232 Agents

Agents are categorized into 16 divisions that mirror real organizational structures, including engineering, design, marketing, product, and QA. Each of the 232 agents has defined roles and expected outputs.

Examples of use cases include Startup MVP creation, Marketing Campaign planning, and Enterprise Feature Development. Agents can operate independently while supporting collaboration.

Division Example Approx. Agents Key Deliverables
engineering ~40 Code reviews, design documents
design ~25 UI/UX proposals, prototypes
marketing ~30 Campaign plans, content ideas
product ~20 Requirements, roadmaps

Source: Official GitHub repository (https://github.com/msitarzewski/agency-agents) as of June 2026.


Real-World Use Cases and Examples

The Agency can be applied to MVP development in startups or marketing campaign planning. By combining multiple agents, complex tasks that are difficult for a single AI can be distributed.

For instance, an engineering agent writes code, a design agent proposes UI, and a marketing agent outlines promotion plans. While tool-specific conversions may be required, the basic workflow is immediately testable.


Considerations and Limitations When Adopting

Installation requires Git and the target AI tool environment. Some tools mandate running conversion scripts. Commercial use is permitted under the MIT license, but human review of agent outputs is recommended.

Note that the agent count has been updated to 232; older references listing 147 should be verified against the latest repository. For security, avoid tasks involving sensitive data.


Frequently Asked Questions (FAQ)

Q1: How many agents does The Agency have?

The official repository currently lists 232 agents across 16 divisions. Check the GitHub page for the latest count.

Q2: Can it be used with tools other than Claude Code?

Yes. It supports Cursor, Gemini CLI, GitHub Copilot, OpenCode, and others via conversion scripts.

Q3: What are the prerequisites for installation?

Git and the CLI environment of the target AI tool are required. install.sh and convert.sh are provided.

Q4: Is commercial use allowed?

Yes, under the MIT license. However, the user remains responsible for final outputs.

Q5: How can I add custom agents?

Create new agent definition files following the template and add them using the installation scripts.


Comparison Table: Supported Tools

Tool Installation Path Conversion Required
Claude Code ~/.claude/agents/ No
Cursor .cursor/rules/ .mdc conversion
Gemini CLI ~/.gemini/agents/ Yes
OpenCode .opencode/agents/ Yes

Source: Official GitHub repository (https://github.com/msitarzewski/agency-agents)


Summary

The Agency represents an effort to elevate AI from a simple assistant to an organizational framework. By leveraging 232 specialized agents, you can streamline development, design, and marketing phases. Start by cloning the official repository and testing it in your own tool environment.

Related new article:

krona23

Author

krona23

Over 20 years in the IT industry, serving as Division Head and CTO at multiple companies running large-scale web services in Japan. Experienced across Windows, iOS, Android, and web development. Currently focused on AI-native transformation. At DevGENT, sharing practical guides on AI code editors, automation tools, and LLMs in three languages.

DevGENT about →

Leave a Reply

Trending

Discover more from DevGENT

Subscribe now to keep reading and get access to the full archive.

Continue reading