Commands Overview

ClaudeKit provides a comprehensive set of slash commands to accelerate your development workflow. Each command is designed for specific tasks and automatically orchestrates the appropriate agents.

Commands → Skills Migration

As of engineer@1.3.0+, many commands have been migrated to skills invoked with / prefix.

What changed:

  • Commands like /cook (was /code), /design/frontend-design, etc.
  • Skills are invoked with / prefix just like commands
  • Cleaner naming and better workflow organization
  • Better token efficiency

See: Migration Guide for complete details and workflow comparisons.

Command Categories

Core Development

  • /ask - Ask questions about the codebase
  • /bootstrap - Initialize new projects with spec-driven development
  • /ck-help - Get help about ClaudeKit commands
  • /coding-level - Set coding complexity level
  • /journal - Write development journal entries
  • /kanban - Manage project tasks in Kanban board
  • /preview - Preview current work
  • /test - Run test suite and get results
  • /use-mcp - Use MCP server tools
  • /watzup - Get project status and recent changes
  • /worktree - Manage git worktrees
  • /wt - Shorthand for /worktree

Note: /cook and /debug are now skills. See migration notice above.

Planning

  • /plan - Create implementation plans
  • /plan
    - Quick planning
  • /plan
    - Detailed planning with research
  • /plan
    - Create plan with 2 approaches
  • /plan
    - Parallel planning with multiple researchers
  • /plan
    - Analyze CI failures and create fix plan
  • /plan
    - Create conversion optimization plan
  • /plan
    - Archive completed plans
  • /plan
    - Validate plan structure

Documentation

  • /docs - Manage project documentation
  • /docs
    - Initialize project documentation
  • /docs
    - Update project documentation
  • /docs
    - Summarize project documentation

Review

Other

Quick Command Reference

Most Used Commands

# Project Setup
/bootstrap [description]         # Initialize new project
/coding-level [level]            # Set coding complexity level

# Feature Development
/plan [feature description]      # Plan the feature
/plan:hard [complex feature]     # Detailed planning with research
/plan:parallel [feature]         # Parallel planning
/cook [implement feature]        # Implement from plan

# Documentation
/docs:init                       # First-time setup
/docs:update                     # After making changes
/docs:summarize                  # Generate documentation summary

# Testing
/test                            # Run test suite
/test:ui                         # Run UI tests
/debug [issue description]       # Debug issues

# Git Workflow
/check-and-commit                # Check quality and commit
/worktree [feature-name]         # Create git worktree

# Project Status
/watzup                          # What's the current state?
/ask [question]                  # Ask about codebase
/kanban                          # View project dashboard

Command Syntax

Basic Syntax

/command [required-argument] [optional-argument]

Examples

# No arguments
/test
/watzup
/docs:init

# Required argument
/plan [add user authentication]
/ask [how does routing work?]
/bootstrap [create REST API]

# Optional arguments
/git pr                          # PR to default branch
/git pr [develop]                # PR to develop
/git pr [main] [feature-branch]  # PR from feature to main

# Multiple arguments
/plan:parallel [feature] [3]     # Parallel planning with 3 researchers

Command Workflows

Starting a New Project

1. /bootstrap [project description]
   # OR
   ck init --kit engineer

2. # Customize requirements through Q&A

3. # System automatically:
   - Researches best practices
   - Creates implementation plan
   - Implements features
   - Generates tests
   - Sets up documentation

Developing a Feature

1. /plan [feature description]
   # Creates detailed implementation plan

2. # Review plan in plans/ directory

3. /cook "Implement the feature as planned"
   # Implements based on plan

4. /test
   # Validates implementation

5. /check-and-commit
   # Checks quality and commits

Investigating Issues

# Debug issues
/debug "Debug the login issue - users can't authenticate"
# - Analyzes logs and code
# - Identifies root cause
# - Suggests solutions

# Run tests
/test
# - Runs full test suite
# - Shows coverage report

# UI testing
/test:ui
# - Runs UI/E2E tests
# - Visual regression testing

Updating Documentation

# After implementing features
/docs:update

# When onboarding new team members
/docs:summarize

# When starting with existing codebase
/docs:init

Command Best Practices

Use the Right Command for the Task

Correct Usage

# Simple questions
/ask [how does routing work?]

# Planning with research
/plan:hard [add OAuth2 authentication]

# Quick planning
/plan:fast [update button styling]

# Parallel planning for complex features
/plan:parallel [rebuild authentication system]

Incorrect Usage

# Don't use fast for complex features
/plan:fast [rebuild entire authentication system]

# Don't skip planning for major features
# (implement without /plan first)

Provide Clear Descriptions

Clear

/plan [add OAuth2 authentication with Google and GitHub providers]
"Debug API issue - returns 500 error when creating user with empty email"
/ask [how is user authentication currently implemented?]

Vague

/plan [add auth]
"Something's broken, please fix"
/ask [how does this work?]

Review Before Committing

# 1. Implement feature

# 2. Test
/test

# 3. Review changes
git diff

# 4. Quality check and commit
/check-and-commit

Use Sequential Commands for Complex Tasks

# 1. Understand codebase
/ask [how is authentication currently implemented?]

# 2. Plan changes with research
/plan:hard [migrate from session-based to JWT authentication]

# 3. Review plan
/preview plans/latest-plan.md

# 4. Implement based on plan
/cook "Implement JWT authentication migration as planned"

# 5. Test
/test

# 6. Debug if needed
/debug "Debug the JWT token validation issue"

# 7. Quality check and commit
/check-and-commit

Command Flags and Options

Some commands support flags:

/plan

/plan [feature]              # Intelligent planning
/plan:fast [feature]         # Quick planning without research
/plan:hard [feature]         # Detailed planning with research
/plan:two [feature]          # Two different approaches
/plan:parallel [feature]     # Parallel planning with researchers
/plan:ci [CI-URL]            # Analyze CI failures
/plan:cro [content]          # Conversion optimization

Understanding Command Output

Commands provide structured output:

Planning Commands

planner Agent: Analyzing codebase...

Research Results:
- OAuth2 best practices reviewed
- Existing auth patterns identified
- Security considerations documented

Implementation Plan Created:
📄 plans/oauth-implementation.md

Plan Summary:
1. Install dependencies (passport, passport-google-oauth20)
2. Configure OAuth2 providers
3. Implement callback routes
4. Add session management
5. Generate tests
6. Update documentation

Estimated time: 2-3 hours
Files to create: 5
Files to modify: 3

Next: Review plan, then implement

Implementation Output

fullstack-developer Agent: Implementing from plan...

Dependencies Installed:
✓ passport (0.6.0)
✓ passport-google-oauth20 (2.0.0)

Files Created:
✓ src/auth/oauth-config.js
✓ src/auth/google-strategy.js
✓ src/routes/auth-callback.js

Tests Generated:
✓ tests/auth/oauth.test.js (15 tests)

Documentation Updated:
✓ docs/api/authentication.md

Implementation complete!

Next: Run /test to validate

Test Commands

tester Agent: Running test suite...

Test Results:
✓ Unit tests: 45 passed
✓ Integration tests: 12 passed
✓ E2E tests: 8 passed

Coverage: 87.3%

All tests passed!

Next: Review changes, then /git cm

Troubleshooting Commands

Command Not Found

Problem: /command not recognized

Solutions:

  1. Verify you’re in a ClaudeKit project (ls .claude/)
  2. Check command exists (ls .claude/commands/)
  3. Run ck init to get latest commands
  4. Restart Claude Code

Command Fails

Problem: Command errors during execution

Solutions:

  1. Check error message for specific issue
  2. Verify prerequisites (API keys, dependencies)
  3. Review agent logs
  4. Try command with simpler input
  5. Use /debug to investigate

Unexpected Results

Problem: Command doesn’t do what expected

Solutions:

  1. Review command documentation
  2. Check if using correct command for task
  3. Provide more specific description
  4. Review generated plans before implementing
  5. Use feedback to refine

Next Steps

Explore specific command categories:

Or learn about:


Key Takeaway: ClaudeKit commands provide a natural, intuitive interface to powerful agent orchestration, making complex development tasks simple and repeatable.