Skip to main content

Install Components

The CLI supports installing six types of components: agents, commands, MCP servers, hooks, settings, and skills. Each component type serves a specific purpose in enhancing your Claude Code workflow.

Component Installation Syntax

Basic Installation

Install a single component:
cct --agent <agent-name>
cct --command <command-name>
cct --mcp <mcp-name>
cct --hook <hook-name>
cct --setting <setting-name>
cct --skill <skill-name>

Batch Installation

Install multiple components of the same type (comma-separated):
cct --agent frontend-developer,backend-developer,security-auditor
cct --mcp web-fetch,github-integration,filesystem-access
Install different component types together:
cct --agent security-auditor \
    --command security-audit \
    --mcp github-integration \
    --setting read-only-mode

Component Naming Formats

Components support two naming formats:

Direct Name

cct --agent frontend-developer
cct --command setup-testing

Category/Name Format

cct --agent development-team/frontend-developer
cct --command testing/setup-testing
cct --hook automation/simple-notifications
Both formats install to the same flat directory structure.

Agents

What are Agents?

Agents are specialized AI assistants with custom system prompts and behaviors. They provide expert guidance for specific tasks like frontend development, security auditing, or documentation writing.

Installation Location

Agents install to:
.claude/agents/<agent-name>.md

Examples

# Frontend development
cct --agent frontend-developer
cct --agent development-team/react-specialist

# Backend development
cct --agent backend-developer
cct --agent development-team/api-architect

# Security
cct --agent security-auditor
cct --agent security-team/penetration-tester

# DevOps
cct --agent devops-engineer
cct --agent infrastructure-team/kubernetes-expert

# Documentation
cct --agent technical-writer
cct --agent content-team/api-documentation-specialist

Frontend Developer

Expert in React, Vue, Angular, and modern web frameworks

Backend Developer

API design, database optimization, and server-side logic

Security Auditor

Code security review, vulnerability detection, best practices

DevOps Engineer

CI/CD, infrastructure, deployment automation

Usage After Installation

Invoke agents in Claude Code:
@frontend-developer Help me optimize this React component
@security-auditor Review this authentication code

Commands

What are Commands?

Slash commands provide quick access to common workflows and operations. They accept arguments and can include complex instructions.

Installation Location

Commands install to:
.claude/commands/<command-name>.md

Examples

# Testing
cct --command setup-testing
cct --command testing/generate-tests

# Code review
cct --command code-review
cct --command review/security-check

# Documentation
cct --command generate-docs
cct --command documentation/api-reference

# Deployment
cct --command deploy-staging
cct --command deployment/production-checklist

Command Structure

Commands use the $ARGUMENTS placeholder for user input:
# Command: setup-testing

Setup comprehensive testing infrastructure for $ARGUMENTS.

Include:
- Test framework configuration
- Example test files
- CI/CD integration

Usage After Installation

/setup-testing React components
/generate-docs API endpoints
/code-review authentication module

MCP Servers

What are MCP Servers?

MCP (Model Context Protocol) servers provide Claude Code with access to external services, databases, APIs, and tools.

Installation Location

MCP configurations merge into:
.mcp.json

Examples

# Web access
cct --mcp web-fetch

# File system
cct --mcp filesystem-access

# Version control
cct --mcp github-integration

# Databases
cct --mcp mysql-integration
cct --mcp postgresql-integration

# Memory
cct --mcp memory-integration

# Code analysis
cct --mcp deepgraph-react
cct --mcp deepgraph-typescript
cct --mcp deepgraph-vue

Configuration Merging

When installing MCPs:
  1. Existing .mcp.json is preserved
  2. New servers are added to mcpServers object
  3. Conflicting server names are overwritten
  4. Description fields are automatically removed

MCP Configuration Example

After installing web-fetch:
{
  "mcpServers": {
    "fetch": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-fetch"]
    }
  }
}

Hooks

What are Hooks?

Hooks are automation triggers that execute commands at specific Claude Code lifecycle events.

Installation Location

Hooks merge into:
.claude/settings.json

Hook Types

  • PreToolUse - Execute before tool usage
  • PostToolUse - Execute after tool usage
  • Stop - Execute when session stops
  • Notification - Execute for notifications

Examples

# Git automation
cct --hook automation/git-commit-validation
cct --hook automation/git-auto-commit

# Notifications
cct --hook automation/simple-notifications
cct --hook automation/slack-notifications

# Testing
cct --hook automation/run-tests-before-commit

# Security
cct --hook automation/security-scan

Hook Configuration Example

{
  "hooks": {
    "PreToolUse": [
      {
        "command": "npm run lint",
        "description": "Run linter before file modifications"
      }
    ],
    "PostToolUse": [
      {
        "command": "git add .",
        "description": "Stage changes automatically"
      }
    ]
  }
}

Settings

What are Settings?

Settings are pre-configured Claude Code settings for permissions, models, telemetry, and behavior.

Installation Location

Settings merge into:
.claude/settings.json

Categories

Permissions

cct --setting permissions/allow-npm-commands
cct --setting permissions/deny-sensitive-files
cct --setting permissions/additional-directories

Model Selection

cct --setting model/use-sonnet
cct --setting model/use-haiku
cct --setting model/use-opus

Telemetry

cct --setting telemetry/enable-telemetry
cct --setting telemetry/disable-telemetry

Retention

cct --setting retention/retention-7-days
cct --setting retention/retention-30-days
cct --setting retention/retention-90-days

Security

cct --setting security/read-only-mode
cct --setting security/strict-permissions

Statuslines

cct --setting statusline/context-monitor
cct --setting statusline/token-counter

Statusline Settings

Statusline settings include Python scripts that are automatically downloaded:
cct --setting statusline/context-monitor
This installs:
  • .claude/settings.json (statusline configuration)
  • .claude/scripts/context-monitor.py (Python script)

Settings Configuration Example

{
  "model": "claude-sonnet-4-20250514",
  "permissions": {
    "allow": [
      {
        "type": "exec",
        "pattern": "npm"
      }
    ],
    "deny": [
      {
        "type": "read",
        "pattern": "**/.env*"
      }
    ]
  },
  "telemetryEnabled": false,
  "cleanupPeriodDays": 7
}

Skills

What are Skills?

Skills are specialized workflows and capabilities that extend Claude Code functionality.

Examples

# Web development
cct --skill web-development/roier-seo

# Document processing
cct --skill document-processing/pptx

# Creative design
cct --skill creative-design/algorithmic-art

# Utilities
cct --skill utilities/playwright-skill

Installation Options

Target Directory

Install to a specific directory:
cct --agent frontend-developer --directory ./my-project
cct -d ./my-project --command setup-testing

Dry Run

Preview what would be installed without making changes:
cct --agent security-auditor --dry-run

Silent Mode

Suppress installation output (used in batch operations):
cct --agent frontend-developer --silent

With Prompt Execution

Install components and execute a prompt:
cct --agent security-auditor --prompt "Review authentication code"

Component Discovery

Browse Components

Explore all available components:
  • Website: https://aitmpl.com/
  • Components by category
  • Search and filter functionality
  • Installation copy-paste commands

Not Found Handling

If a component isn’t found, the CLI shows available options:
cct --agent nonexistent-agent
# ❌ Agent "nonexistent-agent" not found
# πŸ“‹ Available Agents:
# ...

Workflow Installation

Install from Workflow Hash

Install a pre-defined workflow:
cct --workflow #security-audit-workflow

Install with YAML Workflow

Provide a base64-encoded YAML workflow:
cct --workflow <base64-yaml> --agent security-auditor

Troubleshooting

Component Not Found

  1. Verify component name spelling
  2. Check component exists at https://aitmpl.com/
  3. Try category/name format: development-team/frontend-developer

Installation Failed

  1. Check network connectivity
  2. Verify target directory permissions
  3. Try with --verbose flag for detailed output

Merge Conflicts

  • MCPs: New servers added, conflicts overwrite existing
  • Settings: Deep merge, new keys added, conflicts overwrite
  • Hooks: Arrays merged, duplicates may occur

Permission Issues

# Ensure write permissions
chmod 755 .claude/

# Check directory ownership
ls -la .claude/

Advanced Usage

Batch Installation Script

Create a setup script:
#!/bin/bash
# setup-claude.sh

cct --agent frontend-developer \
    --agent backend-developer \
    --command setup-testing \
    --command code-review \
    --mcp github-integration \
    --mcp web-fetch \
    --setting permissions/allow-npm-commands \
    --setting model/use-sonnet \
    --hook automation/git-commit-validation

echo "βœ… Claude Code setup complete!"

Team Configuration

Share component configurations:
# Document required components
echo "Required Components:" > CLAUDE-SETUP.md
echo "- Agent: frontend-developer" >> CLAUDE-SETUP.md
echo "- Command: setup-testing" >> CLAUDE-SETUP.md
echo "- MCP: github-integration" >> CLAUDE-SETUP.md

# Install from documentation
cct --agent frontend-developer \
    --command setup-testing \
    --mcp github-integration

Next Steps

Build docs developers (and LLMs) love