Contains Studio Agents: A Full AI Department for Claude Code, Ready to Clone
Written by Arshdeep Singh
Claude Code's sub-agent system is one of the most powerful and underused features in AI-assisted development. Instead of one generalist model handling everything, sub-agents let you route specific tasks to specialized agents — each with their own instructions, context, and expertise.
The problem is building those agents from scratch. Defining a useful agent requires deep thought about scope, constraints, tool access, and interaction patterns. For most developers, that's a significant upfront investment before you get any value.
Contains Studio Agents eliminates that investment entirely. It's an open-source collection of 30+ production-ready Claude Code sub-agents, organized as a complete AI department, ready to clone and use immediately.
GitHub: contains-studio/agents
What It Is
Contains Studio Agents is structured as a set of departments — mirroring how a real company is organized. Instead of one "do everything" agent, you have specialists:
- 🎨 Design — brand guardians, UX designers, visual direction
- ⚙️ Engineering — backend architects, DevOps automators, security reviewers
- 📣 Marketing — growth hackers, content strategists, social media specialists
- 📦 Product — roadmap planners, feature prioritizers, user research analysts
- 📋 Project Management — sprint planners, standup facilitators, dependency trackers
- 🏢 Studio Operations — process designers, retrospective facilitators, culture builders
- 🧪 Testing — QA engineers, test case generators, coverage analysts
Each agent in each department is a .md file containing a carefully crafted system prompt, tool guidelines, and behavioral constraints.
Installation
Three commands. Done.
git clone https://github.com/contains-studio/agents
cp -r agents/agents/* ~/.claude/agents/
# Restart Claude Code
After restarting, all 30+ agents are available to Claude Code. No configuration, no API keys, no setup wizard.
The Agents (Selected)
Engineering Department
backend-architect
Designs scalable backend systems. Focuses on API design, data modeling, service boundaries, and avoiding premature optimization. Ask it to review your architecture and it'll push back on the right things.
devops-automator
Specializes in CI/CD pipelines, infrastructure as code, container orchestration, and deployment automation. Knows Terraform, GitHub Actions, Docker, and Kubernetes in depth.
security-reviewer
Reviews code for vulnerabilities, OWASP Top 10, secrets exposure, and dependency risks. Useful for pre-PR security checks without the overhead of a full audit.
Marketing Department
growth-hacker
Experiment-driven marketing strategy. A/B tests, acquisition funnels, retention loops. Talks in metrics, not vibes.
tiktok-strategist
Short-form video content strategy specific to TikTok's algorithm and audience. Knows what hooks work, what formats get completed, what drives follows.
content-strategist
Long-form content planning — blog calendars, SEO strategy, thought leadership positioning. Works well paired with the brand-guardian agent.
Product Department
feedback-synthesizer
Takes raw user feedback (support tickets, reviews, interviews) and synthesizes it into structured insights, themes, and prioritized feature signals. Removes the bias from manual reading.
roadmap-planner
Helps prioritize features using frameworks like RICE, ICE, and value/effort matrices. Useful when you have 50 backlog items and need to decide what actually ships next.
Design Department
brand-guardian
Enforces brand consistency — voice, tone, visual language, messaging hierarchy. Useful for reviewing copy, designs, or anything that represents your company externally.
How It Works in Practice
The elegance of the sub-agent system is that routing is automatic. You don't need to specify which agent handles your request — Claude Code reads your task description and delegates to the most appropriate specialist.
Example interactions:
"Review this PR for security vulnerabilities" →
security-reviewerhandles it"We have 40 backlog items. Help me decide what ships in Q2." →
roadmap-plannertakes over"Write three TikTok scripts for our product launch" →
tiktok-strategistactivates"Our pipeline is taking 20 minutes to build. Find the bottleneck." →
devops-automatordigs in
You describe the work in natural language. The right expert shows up.
Why This Matters for AI-Assisted Work
There's a meaningful difference between using AI as a general assistant and using AI through a system of specialists.
Generalist AI agents tend to give generalist answers. When you ask a single agent to review your architecture, write marketing copy, and plan your testing strategy in the same session, you get reasonable outputs — but nothing that reflects deep domain expertise.
Specialist agents behave differently. A backend-architect agent that's been given focused context about system design patterns, tradeoffs, and anti-patterns will produce meaningfully better architecture reviews than a generalist agent asked to "think like a senior backend engineer."
Contains Studio Agents is a practical demonstration of this. It's not theoretical. You can install it in 3 commands and immediately experience the difference between asking a generalist "review this Terraform" and asking a devops-automator who's been specifically prompted on IaC best practices.
Extending the Collection
The format is intentionally simple. Each agent is a single markdown file:
# Agent Name
## Role
[Brief description of the agent's specialty]
## Responsibilities
[What this agent does]
## Guidelines
[How this agent thinks and behaves]
## Tools
[What tools this agent uses]
Adding a custom agent is as simple as writing a new .md file and dropping it in ~/.claude/agents/. The Contains Studio repo gives you a clear template and a high-quality reference set to build from.
Who Should Use This
- Solo developers who want specialist AI reviewers without a team
- Startups that need to move fast across engineering, product, and marketing
- Claude Code users looking to get significantly more value from sub-agents
- Anyone building their own agent collection and wanting a strong starting point
Final Thoughts
Contains Studio Agents represents a mature approach to AI-assisted work: not one smart assistant, but a coordinated team of specialists. The open-source release makes that team available to everyone.
Three commands. Thirty specialists. Start delegating.
👉 github.com/contains-studio/agents
Written by Arshdeep Singh
Top comments (0)