Skip to content

SLASH_COMMANDS_GUIDE.md

Latest commit

 

History

History
336 lines (276 loc) · 7.74 KB

SLASH_COMMANDS_GUIDE.md

File metadata and controls

336 lines (276 loc) · 7.74 KB

Slash Commands Guide

Overview

Clode Studio provides both built-in slash commands and a visual Command Studio for creating custom commands. Commands enhance your workflow by providing quick access to common actions and custom prompts.

Built-in Commands

File Commands

  • /open [file] - Open a file in the editor (aliases: /o, /edit)
  • /save - Save the current file (alias: /s)
  • /close - Close the current file
  • /find <pattern> - Search in files (aliases: /search, /grep)

Task Management

  • /task <description> - Create a new task (aliases: /todo, /t)
  • /tasks - Show all tasks (alias: /todos)

Context Management

  • /context - Show context usage (alias: /ctx)
  • /optimize - Optimize context usage
  • /checkpoint [name] - Create a context checkpoint (alias: /cp)

Knowledge Base

  • /knowledge [query] - Search knowledge base (aliases: /kb, /know)
  • /knowledge-add <title> - Add new knowledge entry (alias: /kb-add)

Claude Management

  • /claude [name] - Create new Claude instance (alias: /new)
  • /personality <name> - Set Claude personality (alias: /p)
  • /session - Resume a previous session (alias: /resume)
  • /think [level] - Set thinking level: normal, more, hard, ultra

System Commands

  • /memory - Edit CLAUDE.md memory files (aliases: /mem, /claude.md)
  • /mcp - Show MCP connections
  • /commands - Open command studio (aliases: /cmd, /slash)
  • /help - Show available commands (aliases: /?, /h)
  • /clear - Clear terminal/chat (alias: /cls)
  • /reload - Reload the application

Workspace

  • /workspace - Switch workspace (alias: /ws)

Custom Commands

Command Studio

Access the Command Studio:

  1. Click the Commands tab in the bottom panel
  2. Or use /commands in Claude
  3. Or press Cmd/Ctrl + Shift + C (coming soon)

Creating Custom Commands

1. Basic Command

name: review
description: Review code for best practices
argument-hint: <file or directory>
---
Review the code in {{ args }} and provide feedback on:
1. Code quality and best practices
2. Potential bugs or issues
3. Performance considerations

2. Command with Tools

name: test
description: Generate tests for code
allowed-tools:
  - read_file
  - write_file
---
Generate comprehensive unit tests for {{ args }}.
Use the appropriate testing framework.

3. Command with Bash Execution

name: status
description: Show git and project status
---
!git status --short
!npm list --depth=0

Analyze the above status and suggest next steps.

4. Command with File Inclusion

name: review-pr
description: Review pull request changes
---
@.git/COMMIT_EDITMSG
!git diff --cached

Review these changes for the pull request.

Command Syntax

Placeholders

  • {{ args }} - All arguments passed to the command
  • {{ arg1 }}, {{ arg2 }} - Individual arguments
  • $1, $2 - Alternative syntax for arguments

Special Operators

  • !command - Execute bash command and include output
  • @file.txt - Include contents of a file
  • {{ VAR }} - Environment variable substitution

File Organization

Commands are stored in:

  • Project: .claude/commands/ in your project
  • Personal: ~/.claude/commands/ in home directory

Categories via Subdirectories

.claude/commands/
├── dev/
│   ├── review.md
│   └── test.md
├── git/
│   ├── commit.md
│   └── pr.md
└── docs/
    └── generate.md

Use commands with categories: /git/commit, /dev/review

Advanced Examples

1. Smart Commit Command

name: commit
description: Create a smart git commit
allowed-tools:
  - run_bash_command
---
!git diff --cached
!git status --short

Based on the changes above, create a conventional commit message.
Then use run_bash_command to commit with that message.

2. API Documentation Generator

name: api-docs
description: Generate API documentation
argument-hint: <controller-file>
allowed-tools:
  - read_file
  - write_file
---
Read {{ args }} and generate OpenAPI documentation.
Consider:
- All endpoints
- Request/response schemas
- Authentication requirements
- Error responses

Save the documentation to docs/api/{{ arg1 }}.yml

3. Refactoring Assistant

name: refactor
description: Suggest refactoring improvements
argument-hint: <file>
allowed-tools:
  - read_file
---
Analyze {{ args }} for refactoring opportunities:

1. Code smells
2. Design pattern applications
3. Performance improvements
4. Testability enhancements

Provide specific before/after examples.

4. Security Audit

name: security
description: Security audit for code
allowed-tools:
  - read_file
  - search_files
---
!grep -r "password\|secret\|key\|token" . --include="*.js" --include="*.ts"

Perform a security audit:
1. Check for exposed secrets
2. SQL injection vulnerabilities
3. XSS possibilities
4. Authentication bypasses
5. Dependency vulnerabilities

Best Practices

Naming Conventions

  • Use lowercase with hyphens: generate-docs
  • Be descriptive but concise
  • Use categories for organization
  • Avoid conflicts with built-in commands

Writing Templates

  1. Start with clear instructions
  2. Use markdown formatting
  3. Specify expected output format
  4. Include examples when helpful
  5. List edge cases to consider

Tool Permissions

  • Only request needed tools
  • Be specific about file access
  • Consider read-only vs write access
  • Document why tools are needed

Performance

  • Avoid expensive bash commands
  • Cache results when possible
  • Use specific file paths
  • Limit search scopes

Command Discovery

Listing Commands

  • Type / to see suggestions
  • Use /help for full list
  • Check Command Studio for visual browser
  • Project commands override personal ones

Search and Filter

In Command Studio:

  • Search by name or description
  • Filter by category
  • View by source (project/personal)
  • See allowed tools at a glance

Sharing Commands

Export/Import

  1. Commands are just markdown files
  2. Share via git repositories
  3. Copy to .claude/commands/
  4. Maintain in version control

Team Commands

  1. Store in project's .claude/commands/
  2. Commit to repository
  3. Document in README
  4. Use categories for organization

Troubleshooting

Command Not Found

  1. Check spelling and case
  2. Verify file exists in commands directory
  3. Ensure .md extension
  4. Check for category prefix

Command Not Working

  1. Test in Command Studio
  2. Verify placeholder syntax
  3. Check tool permissions
  4. Test bash commands separately

Performance Issues

  1. Simplify complex commands
  2. Reduce file operations
  3. Optimize bash commands
  4. Use specific file paths

Tips and Tricks

1. Command Aliases

Create shorter versions:

name: gp
description: Git pull and update
---
!git pull
!npm install
Update complete. Check for any conflicts.

2. Interactive Commands

Ask for clarification:

name: create
description: Create a new component
---
What type of component would you like to create?
1. React functional component
2. Vue composition component
3. Web component

Please specify any props or features needed.

3. Chained Commands

Combine multiple operations:

name: deploy
description: Full deployment process
---
!npm test
!npm run build
!git status

If all tests pass and build succeeds, 
guide me through deployment steps.

4. Context-Aware Commands

Use current context:

name: explain
description: Explain current context
---
Based on our current conversation and the files
we've been working with, provide a summary of:
1. What we're trying to accomplish
2. Progress so far
3. Next steps

Remember: Custom commands are powerful tools for encoding your workflow. Start simple and evolve them as you discover patterns in your work!