Commands are custom slash commands that provide Claude Code with specialized workflows, automation, and context-aware operations. They’re like custom macros that execute complex tasks with a simple /command syntax.
What Are Commands? Commands are markdown files (.md) stored in .claude/commands/ that define:
Workflow steps - Automated sequences of operationsContext gathering - What information to collect before executionTool usage - Which Claude Code tools the command needsOutput format - How results are presented
Commands appear in Claude Code’s command palette (press /) and provide autocomplete with hints.
Command Structure Commands use YAML frontmatter followed by markdown instructions:
--- allowed-tools : Read, Write, Edit, Bash argument-hint : [ pipeline-name ] | setup | status | fix description : Manage and automate CI/CD pipeline configuration with GitHub Actions --- # CI/CD Pipeline Manager Manage CI/CD pipeline automation: $ARGUMENTS ## Current Pipeline State - GitHub Actions: ! `find .github/workflows -name "*.yml"` - CI configuration: @.github/workflows/ - Package scripts: @package.json ## Task Automate CI/CD pipeline management with comprehensive workflow orchestration. ## Pipeline Operations ### Setup New Pipeline Create complete CI/CD pipeline with: ...
Frontmatter Fields Field Required Description allowed-toolsYes Comma-separated list of tools command can use argument-hintNo Usage hint shown in autocomplete descriptionYes Clear description of command’s purpose
Special Syntax Commands support special syntax for dynamic context:
$ARGUMENTS - User-provided arguments after command name@path/to/file - Read and include file contents!\command“ - Execute shell command and include output
Command Categories Setup & Configuration
Git Workflow
Deployment
Database
Security
Documentation
Project setup and configuration commands:
setup-testing - Configure testing framework (Jest, Vitest, etc.)setup-linting - Set up ESLint, Prettier, and code quality toolssetup-typescript - Initialize TypeScript with optimal configsetup-docker - Create Docker and docker-compose files npx claude-code-templates@latest --command setup/setup-testing
Usage: /setup-testing jest /setup-linting eslint+prettier
Git operations and workflow automation:
conventional-commits - Format commits with conventional commit stylebranch-strategy - Implement git-flow or trunk-based developmentpr-template - Create pull request templateschangelog-generator - Generate changelog from commits npx claude-code-templates@latest --command git-workflow/conventional-commits
Usage: /conventional-commits feat: add user authentication /changelog-generator v2.0.0
Deployment and infrastructure commands:
ci-pipeline - GitHub Actions CI/CD pipeline setupvercel-deploy - Vercel deployment configurationdocker-deploy - Docker deployment workflowkubernetes-deploy - Kubernetes manifests and deployment npx claude-code-templates@latest --command deployment/ci-pipeline
Usage: /ci-pipeline setup /vercel-deploy production
Database operations and migrations:
migration-create - Create database migration filesseed-data - Generate seed data for testingschema-diff - Compare database schemasquery-optimizer - Analyze and optimize SQL queries npx claude-code-templates@latest --command database/migration-create
Usage: /migration-create add_user_roles /query-optimizer slow-queries.sql
Security analysis and hardening:
vulnerability-scan - Scan dependencies for vulnerabilitiessecurity-audit - Comprehensive security reviewsecrets-check - Find exposed secrets in codepermissions-audit - Review access controls npx claude-code-templates@latest --command security/vulnerability-scan
Usage: /vulnerability-scan /security-audit src/api/
Documentation generation and maintenance:
api-docs - Generate API documentation from codereadme-generator - Create comprehensive README fileschangelog-update - Update changelog with recent changescomponent-docs - Document React/Vue components npx claude-code-templates@latest --command documentation/api-docs
Usage: /api-docs src/api/ /readme-generator
Real-World Example: CI/CD Pipeline Command File : .claude/commands/ci-pipeline.md--- allowed-tools : Read, Write, Edit, Bash argument-hint : [ pipeline-name ] | setup | status | fix description : Manage and automate CI/CD pipeline configuration with GitHub Actions --- # CI/CD Pipeline Manager Manage CI/CD pipeline automation: $ARGUMENTS ## Current Pipeline State - GitHub Actions: ! `find .github/workflows -name "*.yml" -o -name "*.yaml" 2>/dev/null | head -5` - CI configuration: @.github/workflows/ - Package scripts: @package.json - Recent workflow runs: ! `gh run list --limit 5 2>/dev/null || echo "GitHub CLI not available"` ## Task Automate CI/CD pipeline management with comprehensive workflow orchestration. ## Pipeline Operations ### Setup New Pipeline Create complete CI/CD pipeline with: ```yaml name : CI Pipeline on : push : branches : [ main , develop ] pull_request : branches : [ main ] jobs : test : runs-on : ubuntu-latest strategy : matrix : node-version : [ 18 , 20 ] steps : - uses : actions/checkout@v4 - name : Setup Node.js uses : actions/setup-node@v4 with : node-version : ${{ matrix.node-version }} cache : 'npm' - run : npm ci - run : npm run lint - run : npm run test:coverage - run : npm run build
Multi-Environment Deployment Configure staging and production deployments with environment-specific secrets.
Security & Quality Gates
Dependency vulnerability scanning
Secret detection with TruffleHog
SAST analysis with Super Linter
/ci-pipeline setup
/ci-pipeline status
/ci-pipeline fix
## Dynamic Context Gathering Commands can gather context dynamically: ### File References Include file contents in command context: ```markdown Current configuration: @.github/workflows/ci.yml Package scripts: @package.json
When the command runs, Claude reads these files automatically.
Shell Commands Execute commands to gather system information:
Installed dependencies: ! `npm list --depth=0` Git status: ! `git status --short` Recent commits: ! `git log --oneline -5`
Results are included in the command’s context.
User Arguments Capture user input after the command:
Process user request: $ARGUMENTS
Usage: /command arg1 arg2 makes $ARGUMENTS equal to “arg1 arg2”.
Using Commands Installation # Single command npx claude-code-templates@latest --command ci-pipeline # Multiple commands npx claude-code-templates@latest \ --command ci-pipeline \ --command vulnerability-scan \ --command api-docs # With category prefix npx claude-code-templates@latest --command deployment/ci-pipeline
Invocation Commands appear in Claude Code’s command palette:
/ci-pipeline setup /vulnerability-scan /api-docs src/api/
Press / in Claude Code to see all available commands with autocomplete.
Command Best Practices 1. Clear Argument Hints Provide helpful argument hints:
argument-hint : <feature-name> | list | status | rollback
Shows users exactly how to use the command.
2. Context Before Action Gather context before executing:
## Current State - Project type: @package.json - Existing config: @.eslintrc.js - Git status: ! `git status --short`
3. Clear Sections Organize commands into logical sections:
## Current State [Context gathering] ## Task [What the command does] ## Operations [How to execute] ## Examples [Usage examples]
4. Error Handling Include fallbacks for missing tools:
Recent runs: ! `gh run list --limit 5 2>/dev/null || echo "GitHub CLI not available"`
Only request necessary tools:
allowed-tools : Read, Write, Edit # No Bash if not needed
Creating Custom Commands
Create Command File
Create .claude/commands/my-command.md
Add Frontmatter
Define tools, arguments, and description --- allowed-tools : Read, Write, Edit, Bash argument-hint : <action> [options] description : My custom command for specific workflow ---
Define Context
Specify what information to gather: ## Current State - Project files: ! `ls -la` - Config: @config.json
Write Instructions
Provide clear steps for execution: ## Task Execute workflow: $ARGUMENTS ## Steps 1. Analyze current state 2. Execute operation 3. Validate results
Test
Use the command in Claude Code: Command vs Agent When to use commands vs agents:
Use Command Use Agent Specific workflow automation General expertise domain One-time operations Ongoing conversation Context + instructions format Natural language interaction /command syntax@agent syntax
Commands are great for repeatable workflows. Agents are better for exploratory tasks and ongoing assistance.
Advanced Features Conditional Execution Commands can include conditional logic:
If TypeScript project (@tsconfig.json exists): - Use TypeScript configuration Else: - Use JavaScript configuration
Combine tools for complex operations:
1. Read current config (Read) 2. Generate new config (Edit) 3. Install dependencies (Bash) 4. Validate setup (Bash)
Integration with Hooks Commands can be triggered by hooks:
{ "hooks" : { "PostToolUse" : [{ "matcher" : "Edit" , "hooks" : [{ "type" : "command" , "command" : "/validate-code" }] }] } }
Next Steps
Browse Commands Explore 225+ available commands
Create Custom Commands Build your own commands
Hooks Automate command execution
Workflows Combine commands into workflows