Claude Code Market Documentation
Claude Code Market documentation helps you install, use, and customize AI agents for Claude Code. This comprehensive guide covers everything from installing your first agent in under 2 minutes to creating and submitting your own specialized agents to the marketplace.
Step-by-step guides for developers new to Claude Code, experienced developers contributing agents, and teams implementing agent-powered workflows.
Getting Started
Install and use your first agent in under 2 minutes. No configuration required—just download, move, and reference in your Claude Code conversations.
Prerequisites
- →Claude Code CLI installed (download here)
- →Active Claude API subscription ($20-50/month)
- →Terminal access (macOS, Linux, or WSL on Windows)
# Create agents directory
mkdir -p ~/.claude/agents
# Download an agent
curl -o ~/.claude/agents/frontend-architect.md \
"https://ccmarketplace.com/agents/frontend-architect/download"
# Start using in Claude Code
claude code "Use frontend-architect to build a responsive navbar"Why use agents? Agents reduce development time by 60-80% by providing specialized expertise, automated patterns, and production-ready code templates. They eliminate repetitive boilerplate while applying industry best practices automatically. See tutorials for real-world examples.
Installation Guide
Step-by-step instructions for installing agents from CC Marketplace. Works on macOS, Linux, and Windows (WSL).
Find an Agent
# Browse agents at ccmarketplace.com/agentsExplore our marketplace and find an agent that matches your workflow. Filter by category (Frontend, Backend, Testing, DevOps, AI/ML) or search for specific capabilities.
Download Agent File
curl -o agent.md "https://ccmarketplace.com/agents/[agent-slug]/download"Click the download button or use curl to download the agent markdown file. Each agent is a single .md file containing its prompt and configuration.
Move to Claude Directory
mv agent.md ~/.claude/agents/Place the agent file in your Claude Code agents directory. Create the directory if it doesn't exist yet with: mkdir -p ~/.claude/agents
Start Using
# In Claude Code:
"Use the frontend-architect agent to build a navbar"Reference the agent in your Claude Code conversations. The agent's specialized knowledge will enhance Claude's responses for your specific task.
Core Concepts
Understand how Claude Code agents work and how to use them effectively for maximum productivity.
What are Agents?
Agents are specialized prompt files that extend Claude Code with domain expertise. Think of them as expert consultants that Claude can reference for specific tasks. Each agent contains curated knowledge, patterns, and best practices.
How Agents Work
When you reference an agent, Claude Code loads its context and applies its specialized knowledge to your task. Agents contain patterns, best practices, and workflows that guide Claude's code generation.
Agent Types
Agents span all development needs: frontend frameworks (React, Vue), backend services (Node.js, Python), testing strategies (Jest, Playwright), DevOps automation (Docker, CI/CD), database design (SQL, NoSQL), and more.
Best Practices
Be specific when invoking agents. Instead of 'build a form', try 'use the react-forms agent to build an accessible multi-step form with validation'. Specificity improves results by 3-5x.
Agent Usage
Quick examples of invoking agents for common tasks. For complete walkthroughs, see our tutorials.
Frontend Development
Generate React components with TypeScript and Tailwind
"Use frontend-architect to build a responsive product card with hover animations and TypeScript types"API Development
Create REST endpoints with validation and documentation
"Use backend-pro to create a user authentication API with JWT tokens and rate limiting"Testing Automation
Generate comprehensive test suites automatically
"Use test-automator to create E2E tests for the checkout flow with Playwright"Learn by example: Visit our tutorials section for complete step-by-step guides building real applications with agents.
Creating Agents
Build and submit your own specialized agents to CC Marketplace. Share your expertise with the community and help developers ship faster.
Define Agent Scope
Choose a specific domain where you have expertise. Best agents solve focused problems (React forms, REST APIs, Docker deployment) rather than broad categories.
Write Clear Instructions
Document when to use the agent, what it generates, and key patterns it follows. Include examples of good inputs and expected outputs.
Add Code Examples
Include 3-5 code examples showing best practices, common patterns, and edge cases. Examples teach agents your preferred coding style.
Test Thoroughly
Test your agent with various prompts. Verify it handles different scenarios, edge cases, and follows your documented patterns consistently.
Submit to Marketplace
Use the agent submission form with a clear description, category, and tags. Include what problems it solves and who benefits from using it.
Agent Template
# Agent Name
## Purpose
Clear description of what this agent does and when to use it.
## Usage
How to invoke this agent in Claude Code conversations.
## Best Practices
- Best practice 1
- Best practice 2
- Best practice 3
## Code Patterns
### Pattern 1: [Pattern Name]
```typescript
// Example code showing the pattern
```
### Pattern 2: [Pattern Name]
```typescript
// Another example
```
## Common Use Cases
1. Use case 1 with example
2. Use case 2 with example
## Output Expectations
What developers should expect when using this agent.Submission Guidelines
- Focus: Solve a specific problem well rather than trying to cover everything
- Examples: Include 3-5 code examples showing best practices and common patterns
- Documentation: Clear instructions on when and how to use the agent
- Testing: Test with various prompts before submission
- Originality: Ensure your agent provides unique value not covered by existing agents
Quick Reference
Essential guidelines for effective agent usage. For detailed tutorials and examples, see our tutorials section.
Be Specific
Specific prompts yield 3-5x better results.
"Use frontend-architect to build an accessible button with loading state"One Agent Per Task
Focus on one agent at a time for best results. Switch agents when changing domains (frontend → backend → testing).
Provide Context
Share tech stack, existing patterns, and project structure. Agents perform better with context about your codebase.
Review Generated Code
Always review for security, performance, and alignment with your architecture before deploying.
Want detailed guidance? Check out our tutorials section for step-by-step guides on building with agents, advanced techniques, and real-world examples.
Troubleshooting
Common issues and their solutions. Get unstuck quickly with these frequently reported problems and fixes.
Agent not found
installationVerify the agent file is in ~/.claude/agents/ with a .md extension. Check file permissions with: ls -la ~/.claude/agents/. Ensure the directory exists: mkdir -p ~/.claude/agents/
Agent not being applied
usageExplicitly reference the agent name in your prompt. Example: "Use the [agent-name] to build...". Check that Claude Code is using the latest version and restart if needed.
Poor quality output
qualityBe more specific in your instructions. Provide context about your project structure, tech stack, and requirements. Include examples of what you want. Add follow-up prompts to refine output.
Conflicting agent advice
usageUse one agent at a time for focused tasks. Different agents may have different approaches or best practices. Clear conversation context when switching domains.
Slow response times
performanceLarge agents with extensive examples may increase response time. Consider breaking complex requests into smaller, focused tasks. Check your Claude API rate limits.
Agent generates outdated code
qualityCheck for agent updates on Claude Code Market. Download the latest version if available. Some agents may reference older library versions—specify versions in your prompt.
Can't customize agent behavior
customizationEdit the agent markdown file directly in ~/.claude/agents/. Add your preferences, coding standards, or project-specific patterns to the agent instructions.
Agent not understanding project context
usageProvide more context in your prompt: tech stack, existing patterns, file structure. Reference specific files: "Following the pattern in /components/Button.tsx..."
Still need help?
Check the API reference for advanced usage or contact our support team.
Frequently Asked Questions
Quick answers to common questions about Claude Code agents, CC Marketplace, and agent usage.
How do I install a Claude Code agent?
Download the agent markdown file from Claude Code Market, move it to ~/.claude/agents/ directory, and reference the agent name in your Claude Code conversations. Installation takes under 2 minutes with no configuration required.
What are Claude Code agents?
Agents are specialized prompt files that extend Claude Code with domain expertise. Each agent contains patterns, best practices, and workflows for specific development tasks like frontend architecture, backend APIs, or testing automation.
How do agents improve development speed?
Agents reduce development time by 60-80% by providing specialized knowledge, automated code generation, built-in testing patterns, and production-ready templates. They eliminate repetitive boilerplate and apply industry best practices automatically.
Can I use multiple agents simultaneously?
Yes, but it's recommended to use one agent per focused task for best results. You can switch between agents for different parts of your project (e.g., frontend-architect for UI, backend-pro for APIs).
How do I create my own agent?
Create a markdown file with clear instructions, examples, and best practices for your domain. Include context about when to use the agent, code patterns, and expected outputs. Submit via the agent submission form for marketplace inclusion.
Are agents free to use?
Yes, all agents in Claude Code Market are free to download and use. You need an active Claude API subscription ($20-50/month) to run Claude Code and use agents.
What's the difference between agents and plugins?
Agents are prompt-based extensions that guide Claude's behavior, while plugins execute code. Agents require no installation beyond placing a file in a directory and work within Claude's native capabilities.
How do I update an agent?
Download the latest version from Claude Code Market and replace the existing file in ~/.claude/agents/. Agent updates typically include improved prompts, new patterns, or expanded capabilities based on community feedback.
Can agents access my codebase?
No, agents don't access your code directly. Claude Code reads your codebase and applies agent instructions when you reference the agent in your conversation. Agents guide Claude's responses but don't execute code.
Why isn't my agent working?
Check that: 1) The agent file is in ~/.claude/agents/ with .md extension, 2) You're explicitly referencing the agent name in your prompt, 3) File permissions allow reading, 4) Claude Code is using the latest version.
How specific should my agent prompts be?
Very specific. Instead of 'build a form', use 'use the react-forms agent to build an accessible multi-step registration form with email validation and loading states'. Specificity dramatically improves output quality.
Can I modify downloaded agents?
Yes, all agents are plain markdown files you can edit. Customize prompts, add project-specific patterns, or adjust outputs to match your coding style. Consider contributing improvements back to the marketplace.