Claude Crew

Pre-built AI agent personas for Claude Code

Drop them in. They work.

Everyone builds the orchestra. We wrote the musicians.

---

What is this?

A collection of ready-to-use AI agent personas for Claude Code. Each agent is a specialized expert with a defined personality, methodology, and output format.

Copy an agent into your project. Point Claude Code at it. Get expert-level output.

Quick Start

# Clone the crew
git clone https://github.com/hlsitechio/claude-crew.git

# Copy agents you need into your project
cp -r claude-crew/agents/code-reviewer ~/.claude/agents/
cp -r claude-crew/fun/gordon-ramsay ~/.claude/agents/

# Use in Claude Code
# Claude will automatically detect agents in .claude/agents/

Or cherry-pick individual agents:

# Just the code reviewer
curl -o .claude/agents/code-reviewer.md \
  https://raw.githubusercontent.com/hlsitechio/claude-crew/main/agents/code-reviewer/agent.md

---

Production Agents

Real tools for real work. Each one is a specialist.

Agent	What it does	Best for
Code Reviewer	Finds bugs, security issues, design problems	PR reviews, code quality
Test Writer	Generates comprehensive tests that catch real bugs	Test coverage, TDD
Doc Generator	Writes docs humans actually read	README, API docs, architecture
Refactorer	Transforms messy code into clean systems	Legacy code, tech debt
API Designer	Designs clean REST/GraphQL APIs	New APIs, API redesign
DevOps	CI/CD, Docker, deployment automation	Infrastructure, shipping
Performance	Finds bottlenecks, optimizes systems	Slow apps, scaling
Accessibility	WCAG compliance, inclusive design	Web apps, compliance
Migrator	Framework/language/architecture migrations	Upgrades, rewrites
Learner	Explains code like a patient teacher	Learning, onboarding
Security Auditor	OWASP Top 10, dependency audit, vulnerability scanning	Security reviews
SysAdmin	Server management, networking, hardening	Infrastructure, ops
Cloud Architect	Cloud infrastructure design, cost optimization	Architecture, scaling
Incident Responder	Production incident management, post-mortems	On-call, outages
Data Engineer	Pipelines, schemas, query optimization	Data systems, ETL

---

Fun Modes

Because coding should be fun. These go viral on Twitter.

Mode	Personality	Sample
Gordon Ramsay	Brutal code reviewer chef	"This function is RAWWW! 200 lines?! SPLIT IT UP!"
Pirate	Nautical developer	"Arrr! A sea monster lurks in yer auth middleware!"
Yoda	Force-wielding code master	"Null pointer, there is. Handle the absence, you must."
Bob Ross	Gentle debugging companion	"Happy little accident on line 47. Let's add a fix right here."
Coach	Motivational hype person	"You just CRUSHED that bug! CHAMPIONSHIP behavior!"
Noir Detective	Hard-boiled bug hunter	"The bug walked in at 3am. Empty catch block. A cover-up."

---

Swarm Templates

Multi-agent team configurations for coordinated work.

Swarm	Agents	Use Case
Full Stack Team	API Designer + Code Reviewer + Test Writer + Doc Generator	Building features end-to-end
Code Review Squad	Code Reviewer + Security Auditor + Performance	Triple-lens code review
Ship Fast Crew	Builder + Test Writer + Doc Generator + DevOps	Maximum velocity shipping

---

How Agents Work in Claude Code

Claude Code supports custom agents through markdown files in your project:

your-project/
├── .claude/
│   └── agents/
│       ├── code-reviewer.md    ← Agent persona
│       ├── test-writer.md      ← Agent persona
│       └── gordon-ramsay.md    ← Fun mode
├── src/
└── ...

When you use the Task tool or subagents, Claude Code reads these agent definitions and adopts the persona, methodology, and output format defined in each file.

Using agents in CLAUDE.md

# CLAUDE.md

## Agents
When asked to review code, use the code-reviewer agent.
When asked to write tests, use the test-writer agent.
When the user types "/fun", switch to gordon-ramsay mode.

---

Contributing

Add your own agents! The crew is always hiring.

1. Fork this repo
2. Create agents/your-agent/agent.md or fun/your-mode/agent.md
3. Follow the format: Personality + Methodology + Output Format + Rules
4. Submit a PR

Agent Template

# Agent Name

[One line: what this agent does]

## Personality
- [How it communicates]
- [What it prioritizes]

## Methodology
[Step-by-step approach]

## Output Format
[How results are structured]

## Rules
- [Non-negotiable behaviors]

---

Related

• claude-memory — Persistent memory for Claude Code sessions
• Claude Code Docs — Official documentation
• MCP Protocol — Model Context Protocol

---

Everyone builds orchestration frameworks. Nobody writes the agents.

We did.

MIT License