A comprehensive collection of specialized AI subagents for Claude Code, designed to enhance development workflows with domain-specific expertise.
This repository contains 36 specialized subagents that extend Claude Code's capabilities. Each subagent is an expert in a specific domain, automatically invoked based on context or explicitly called when needed.
# Clone the repository
git clone https://github.com/yourusername/claude-code-subagents.git
# Copy all subagents to Claude Code's agent directory
cp claude-code-subagents/*.md ~/.claude/agents/
# Restart Claude Code to load the new subagents- backend-architect - Design RESTful APIs, microservice boundaries, and database schemas
- frontend-developer - Build Next.js applications with shadcn/ui, React Server Components, and Tailwind CSS
- mobile-developer - Develop React Native or Flutter apps with native integrations
- graphql-architect - Design GraphQL schemas, resolvers, and federation
- directus-developer - Build and customize Directus applications with extensions, hooks, and API integrations
- drupal-developer - Build and customize Drupal applications with custom modules, themes, and integrations
- python-pro - Write idiomatic Python code with advanced features and optimizations
- golang-pro - Write idiomatic Go code with goroutines, channels, and interfaces
- rust-pro - Write idiomatic Rust code with ownership, lifetimes, and type safety
- typescript-expert - Write type-safe TypeScript with advanced type system features
- devops-troubleshooter - Debug production issues, analyze logs, and fix deployment failures
- deployment-engineer - Configure CI/CD pipelines, Docker containers, and cloud deployments
- cloud-architect - Design AWS/Azure/GCP infrastructure and optimize cloud costs
- database-optimizer - Optimize SQL queries, design efficient indexes, and handle database migrations
- code-reviewer - Expert code review for quality, security, and maintainability
- security-auditor - Review code for vulnerabilities and ensure OWASP compliance
- test-automator - Create comprehensive test suites with unit, integration, and e2e tests
- performance-engineer - Profile applications, optimize bottlenecks, and implement caching strategies
- debugger - Debugging specialist for errors, test failures, and unexpected behavior
- data-scientist - Data analysis expert for SQL queries, BigQuery operations, and data insights
- data-engineer - Build ETL pipelines, data warehouses, and streaming architectures
- ai-engineer - Build LLM applications, RAG systems, and prompt pipelines
- ml-engineer - Implement ML pipelines, model serving, and feature engineering
- api-documenter - Create OpenAPI/Swagger specs and write developer documentation
- payment-integration - Integrate Stripe, PayPal, and payment processors
- quant-analyst - Build financial models, backtest trading strategies, and analyze market data
- legacy-modernizer - Refactor legacy codebases and implement gradual modernization
- accessibility-specialist - Ensure web applications meet WCAG 2.1 AA/AAA standards
- blockchain-developer - Develop smart contracts, DeFi protocols, and Web3 applications
- game-developer - Build games with Unity, Unreal Engine, or web technologies
- ui-ux-designer - Design user interfaces and experiences with modern design principles and accessibility standards
- crypto-trader - Build cryptocurrency trading systems and integrate with exchange APIs
- defi-strategist - Design and implement DeFi yield strategies and liquidity provision
- crypto-analyst - Perform market analysis, on-chain analytics, and sentiment analysis
- arbitrage-bot - Identify and execute crypto arbitrage opportunities across markets
- crypto-risk-manager - Implement risk management systems for crypto portfolios
- Claude Code installed and configured
- Access to the
~/.claude/agents/directory
-
Clone this repository:
git clone https://github.com/yourusername/claude-code-subagents.git cd claude-code-subagents -
Copy subagents to Claude Code's directory:
# macOS/Linux cp *.md ~/.claude/agents/ # Windows copy *.md %USERPROFILE%\.claude\agents\
-
Verify installation:
ls ~/.claude/agents/ -
Restart Claude Code to load the new subagents
Claude Code automatically delegates to the appropriate subagent based on:
- Task context and requirements
- Keywords in your request
- File types being worked on
- Error messages or issues encountered
You can explicitly request a specific subagent:
"Use the code-reviewer to check my recent changes"
"Have the security-auditor review this authentication code"
"Get the performance-engineer to optimize this function"
User: "I just finished implementing the user authentication feature"
Claude: [Automatically invokes code-reviewer after detecting new code]
User: "I'm getting a TypeError in my React component"
Claude: [Automatically invokes debugger to analyze the error]
User: "This query is running slowly"
Claude: [Automatically invokes database-optimizer to analyze and optimize]
User: "Help me build a bot to trade on Binance"
Claude: [Automatically invokes crypto-trader to build trading system]
User: "I want to optimize my yield farming across protocols"
Claude: [Automatically invokes defi-strategist to design yield strategies]
Each subagent follows this structure:
---
name: subagent-name
description: When this subagent should be invoked
tools: tool1, tool2 # Optional - defaults to all tools
---
System prompt defining the subagent's role and capabilities- name: Unique identifier for the subagent
- description: Triggers automatic invocation
- tools: Specific tools the subagent can use (optional)
- system prompt: Detailed instructions and expertise
- Trust automatic delegation - Claude Code knows when to use each specialist
- Be specific when needed - Explicitly mention subagents for targeted help
- Combine specialists - Complex tasks may benefit from multiple experts
- Provide context - More information helps subagents work effectively
- Let subagents focus - Each should excel in one domain
- Write clear descriptions - Help Claude Code know when to invoke
- Include examples - Show how the subagent should respond
- Test thoroughly - Ensure subagents work well together
- Check the installation directory:
ls ~/.claude/agents/ - Ensure files have
.mdextension - Restart Claude Code
- Check file permissions
- Use explicit invocation to test
- Check the subagent's description
- Verify the subagent file is valid Markdown
- Look for syntax errors in the frontmatter
- Avoid overlapping subagent descriptions
- Keep system prompts focused
- Use specific tool restrictions when appropriate
We welcome contributions! To add a new subagent:
-
Create a new file:
subagent-name.md -
Follow the format:
--- name: your-subagent description: Clear description of when to use --- Detailed system prompt...
-
Guidelines:
- Use lowercase, hyphen-separated names
- Write clear, specific descriptions
- Include concrete examples in the prompt
- Test with various scenarios
- Document any special requirements
-
Submit a PR with:
- The new subagent file
- Updated README listing
- Example usage scenarios
Some subagents specify which tools they can use:
---
name: documentation-writer
tools: Read, Write, Edit
---This improves performance and security by limiting capabilities.
Claude Code can use multiple subagents in sequence:
code-revieweridentifies issuesdebuggerinvestigates errorssecurity-auditorchecks vulnerabilities
- Issues: Report bugs or request features
- Discussions: Share use cases and tips
- Contributors: See CONTRIBUTORS.md
- frontend-developer - Now specialized in Next.js 14+, shadcn/ui components, and modern React patterns including Server Components and App Router
- Added 5 specialized crypto trading subagents for trading systems, DeFi strategies, market analysis, arbitrage, and risk management
This collection is provided under the MIT License. See LICENSE for details.
Made with β€οΈ by the Claude Code community