Hire 144 AI Specialists for Free — Agency Agents Crosses 52K GitHub Stars in Just One Week

What if you could hire a marketing expert, a frontend developer, an SEO consultant, and a brand designer all at once inside Claude Code? The open-source project Agency Agents does exactly that. This collection of 144+ AI specialist profiles gained 29,000 new GitHub stars in just one week, pushing its total past 52,000.

So What Exactly Is This?

Agency Agents is a collection of profile files that give AI coding tools a "specialist role." For example, tell Claude Code "Use the frontend developer agent to review this component," and the AI analyzes your code from a frontend expert's perspective. Say "Use the SEO specialist agent to analyze this page," and it responds like a search engine optimization (SEO) pro.

Each profile includes detailed settings for area of expertise, personality, workflow, and success metrics. These aren't simple prompt templates — they're expert simulations refined over months of real-world use.

12 Departments, 144 Specialists — The Full Org Chart

What makes this project impressive is that it's organized like a real company, with distinct departments.

Beyond those, there are departments for Product Management (5), Project Management (6), QA Testing (8), Customer Support (6), Spatial Computing (6), Game Development (19), and Academic Research (5) — plus a specialty department (23) covering everything from blockchain auditing and government consulting to cultural strategy.

Installation — One Copy Command and You're Done

If you're a Claude Code user, setup couldn't be simpler.

# 1. Download the repository
git clone https://github.com/msitarzewski/agency-agents.git

# 2. Copy the agent files to your Claude Code agents folder
cp -r agency-agents/* ~/.claude/agents/

Then just talk to Claude Code like this:

"Use the frontend developer agent to review this component"
"Use the SEO specialist agent to analyze this page's search optimization"
"Use the Reddit community builder agent to create a Reddit marketing strategy for this product"

If you use Cursor, Windsurf, Gemini CLI, Aider, or other tools, an automatic conversion script is provided:

# Auto-detect your tool and convert
./scripts/install.sh

# Or specify a tool directly
./scripts/install.sh --tool cursor

Real-World Use Cases

Startup MVP Development

Activate the frontend developer + backend architect + growth hacker + rapid prototyper + Reality Checker agents simultaneously. Each reviews your code and strategy from their own perspective, giving you the effect of a full team code review — even if you're working solo.

Marketing Campaign Planning

Combine the content creator + Twitter specialist + Instagram curator + Reddit community builder + analytics reporter agents. Since each platform demands a different tone and strategy, you get platform-specific tailored content in one go.

Paid Ad Optimization

Chain agents together: ad audit → tracking setup → PPC (pay-per-click) strategy → keyword analysis → creative strategy → analytics reporting. Work that would cost thousands of dollars at an agency — handled by AI instead.

Why Did It Spread So Fast?

The reasons this project earned 29,000 stars in a single week are clear.

First, the barrier to entry is virtually zero. Just copy the files — no separate installation or configuration required. Second, non-developers can use it immediately. A marketer can use the SEO agent, a designer can use the UX research agent — anyone can instantly create an expert in their own field. Third, the profiles are battle-tested. They were shaped by months of feedback from the Reddit community.

It's MIT-licensed, so commercial use is completely free. Chinese translations already exist in two versions. The project has 44 contributors and over 8,000 forks.

A Word of Caution

Agent profiles are ultimately system prompts (instructions that shape how the AI behaves). They don't change the AI's fundamental capabilities. A marketing agent won't actually run ads or collect data for you. What it does is guide the AI to think and respond from a specific perspective, so the quality of the output still depends on the underlying AI model's performance.

That said, the difference between "just asking AI" and "asking AI with an expert role assigned" is very noticeable in practice — especially for complex decisions that require reviewing multiple perspectives at once.

Related Content — Get Started with AI the Easy Way | Free Learning Guide | More AI News