Claude Code Agents

23 specialized AI agents for software development, infrastructure, and quality assurance.

Overview

Production-ready agents that handle specific development tasks through autonomous delegation. Each agent specializes in a domain (Python, APIs, security, DevOps, etc.) and executes work using the appropriate tools and expertise.

Quick Start

# Install all agents
git clone https://github.com/I-Onlabs/claude-code-agents.git
cp claude-code-agents/*.md ~/.claude/agents/

# Install specific agent
cp claude-code-agents/python-pro.md ~/.claude/agents/

Available Agents

Development

Agent	Specialization	Best For
python-pro	Python 3.11+ expert	Type hints, async/await, data science, Pythonic patterns
backend-developer	API & microservices	FastAPI, REST APIs, database integration, CRUD operations
frontend-expert	Modern web development	React 18+, Next.js 14+, TypeScript, responsive UI
database-architect	Data layer design	Schema design, migrations, query optimization, indexing
api-designer	API architecture	REST/GraphQL design, OpenAPI specs, API versioning

Infrastructure & DevOps

Agent	Specialization	Best For
devops-engineer	Automation & deployment	CI/CD pipelines, Docker, environment configuration
kubernetes-specialist	Container orchestration	K8s deployments, Helm charts, service mesh, autoscaling
data-engineer	Data infrastructure	ETL pipelines, data modeling, warehouse design

Quality & Analysis

Agent	Specialization	Best For
code-reviewer	Quality assessment	Code reviews, best practices, security checks
security-auditor	Security analysis	Vulnerability scanning, auth/authz review, OWASP
qa-expert	Test strategy	Test planning, coverage analysis, QA processes
performance-engineer	Optimization	Profiling, bottleneck analysis, caching strategies
error-detective	Debugging	Root cause analysis, systematic bug investigation

Documentation

Agent	Specialization	Best For
documentation-engineer	Technical writing	API docs, architecture diagrams, developer guides

Orchestration

Agent	Specialization	Best For
task-orchestrator	Multi-task coordination	Breaking down complex features, parallel execution
vibe-coding-coordinator	Preparation phase	Understanding vague requests, gathering requirements
bootstrap-orchestrator	Project initialization	Auto-detecting tech stack, setting up new projects
strategic-vision-architect	Full lifecycle	Zero-to-deployed product development
architectural-cognition-engine	Architecture analysis	Multi-dimensional codebase analysis, tech debt review
research-oracle	External research	Finding docs, analyzing packages, remote code exploration
coordinator-agent	Agent coordination	Multi-agent task management, message bus routing

Problem Solving

Agent	Specialization	Best For
meta-problem-solver	Goal-directed solving	Plan-Act-Observe-Adjust loops, skill/agent creation
problem-solver	Diagnostic investigation	Systematic debugging, multi-source analysis, hypothesis testing
skill-architect	Skill creation	Creating production-ready Claude Code skills and agents

Usage

Automatic (Recommended)

Claude automatically delegates to the right agent based on your request:

You: "Add rate limiting to the API"
→ Delegates to backend-developer

You: "Review this code for security issues"
→ Delegates to security-auditor

You: "Debug why authentication is failing"
→ Delegates to error-detective

You: "Set up CI/CD pipeline"
→ Delegates to devops-engineer

Manual (Optional)

Explicitly specify an agent when needed:

You: "Use python-pro to refactor this function"
You: "Have security-auditor check for vulnerabilities"

Agent Capabilities

Tool Access

Read-Only Agents (Analysis only, cannot modify):

• code-reviewer
• security-auditor
• qa-expert

Full Access Agents (Can implement changes):

• All development agents
• All infrastructure agents
• All orchestration agents

Systematic Methodology

9 agents use ReAct methodology (Think → Act → Observe → Repeat) for systematic problem-solving:

• python-pro, backend-developer, frontend-expert
• database-architect, api-designer
• devops-engineer, kubernetes-specialist
• performance-engineer, error-detective

This prevents rushed implementations and enables self-correction.

File Structure

Each agent is a markdown file with YAML frontmatter:

---
name: agent-name
description: Brief description of capabilities
model: opus|sonnet|haiku
tools: Read, Write, Edit, Bash, Glob, Grep
---

# Agent Content

Detailed instructions, examples, and methodology...

Keeping Updated

# Sync latest changes from local agents
cd claude-code-agents/
./sync-from-local.sh

# Commit and push
git add *.md
git commit -m "Update agents from local"
git push

Contributing

Contributions welcome! When adding or modifying agents:

1. Maintain YAML frontmatter format
2. Provide clear description of capabilities
3. Include examples of when to use
4. Specify appropriate tool access level
5. Test with real scenarios

License

MIT License - See LICENSE for details

---

Repository: 23 specialized agents Categories: Development • Infrastructure • Quality • Documentation • Orchestration • Problem Solving Usage: Automatic delegation via natural language