What Are Slash Commands?

Slash commands are shortcuts that trigger predefined AI workflows:

/review          → Run a comprehensive code review
/test            → Generate tests for the current file
/docs            → Generate documentation
/fix             → Analyze and fix the current error
/refactor        → Suggest and apply refactoring
/security        → Security audit of recent changes

Instead of typing a detailed prompt every time, you type a short command and the agent executes a pre-built workflow.

The benefits:

• Consistent quality (same prompt every time)
• Team standardization (everyone reviews code the same way)
• Speed (one command vs. a paragraph of instructions)
• Composability (commands can call other commands)

Building a /review Command

Create a code review command with a SKILL.md file:

# /review — Code Review

## System Prompt
You are a senior software engineer performing a thorough code review.
Focus on bugs, security, performance, and maintainability.

## Steps
1. Read the git diff of staged or recent changes
2. For each changed file, analyze:
   - Logic errors and potential bugs
   - Security vulnerabilities (OWASP Top 10)
   - Performance concerns
   - Code style and readability
3. Rate overall quality (1-10)
4. List specific issues with severity

## Output Format
### Code Review Summary
- **Quality Score:** X/10
- **Files Reviewed:** N

### Issues Found
For each issue:
- **File:** path/to/file.js:LINE
- **Severity:** Critical | High | Medium | Low
- **Issue:** Description
- **Fix:** Suggested solution

### Positive Highlights
- What's done well (acknowledge good code!)

Building a /test Command

Automated test generation:

# /test — Generate Tests

## System Prompt
You are a testing specialist. Generate comprehensive tests
that cover happy paths, error cases, and edge cases.

## Steps
1. Read the target file (current file or specified file)
2. Identify all public functions and methods
3. For each function, generate tests for:
   - Happy path (normal input → expected output)
   - Error cases (invalid input → appropriate error)
   - Edge cases (empty, null, boundary values)
   - Integration points (mocked dependencies)
4. Use the project's testing framework (read AGENTS.md)
5. Run the tests and fix any failures

## Constraints
- Match existing test patterns in the project
- Use descriptive test names: "should [expected behavior] when [condition]"
- Mock external dependencies, never call real APIs
- Aim for 90%+ coverage of the target file

Building a /docs Command

Documentation that stays in sync:

# /docs — Generate Documentation

## System Prompt
You are a technical writer who values clarity and accuracy.

## Modes
### Mode 1: Single file (default)
- Read the current file
- Generate JSDoc/TSDoc for all exported functions
- Add module-level documentation

### Mode 2: API reference (/docs api)
- Scan all route files
- Generate endpoint documentation
- Include request/response schemas and examples

### Mode 3: Architecture (/docs arch)
- Analyze project structure
- Generate architecture overview
- Include dependency diagram (as ASCII art)

## Output Rules
- Keep descriptions under 2 sentences
- Always include @param, @returns, @throws
- Add @example with realistic usage
- Match existing documentation style in the project

SKILL.md Plugin System

Skills are reusable, shareable prompt packages:

Directory structure:

.claude/skills/
  review/
    SKILL.md       ← The skill definition
  test/
    SKILL.md
  docs/
    SKILL.md
  security-audit/
    SKILL.md
  deploy-check/
    SKILL.md

SKILL.md anatomy:

name: Security Audit command: /security description: Run a security audit on recent changes author: your-team version: 1.2.0

Security Audit

Trigger

Run when: manually invoked or on PR creation

System Prompt

[Detailed instructions for the AI]

Tools Required

• File system (read source code)
• Terminal (run dependency audit)
• Git (read recent changes)

Output

[Expected format] ```

Composing Commands

Commands become powerful when they chain together:

The /ship command — a complete pre-merge workflow:

# /ship — Pre-Merge Checklist

## Steps
1. Run /review on all changed files
   - If quality score < 7: stop and report issues
2. Run /test to generate missing tests
   - If coverage < 80%: stop and report gaps
3. Run /security on all changes
   - If any Critical issues: stop immediately
4. Run /docs to update documentation
5. If all checks pass:
   - Stage all changes
   - Create a commit with a descriptive message
   - Report: "Ready to merge! ✓ Review ✓ Tests ✓ Security ✓ Docs"

This single command replaces a 30-minute manual checklist.

Sharing Skills Across Teams

Skills become a team knowledge base:

Team-wide skill repository:

company-ai-skills/           ← Shared Git repo
  code-review/SKILL.md       ← Standardized reviews
  test-generation/SKILL.md   ← Test standards
  incident-response/SKILL.md ← On-call playbook
  onboarding/SKILL.md        ← New developer setup
  deploy/SKILL.md            ← Deployment checklist

Benefits:

• New team members get expert-level AI tools immediately
• Best practices are encoded, not just documented
• Updates propagate to everyone automatically
• Knowledge doesn't leave when people do

Installation:

# Clone team skills into your project
git clone company-ai-skills .claude/skills

Skills are the bridge between individual AI expertise and organizational knowledge. One person discovers the best prompt for code review — the whole team benefits.

---quiz question: What is the main advantage of slash commands over typing prompts manually? options:

• { text: "They use fewer API tokens", correct: false }
• { text: "Consistent quality, team standardization, and speed — the same optimized workflow every time", correct: true }
• { text: "They work without an internet connection", correct: false } feedback: Slash commands encode your best prompts into reusable workflows. Every team member gets the same quality review, the same thorough tests, and the same documentation standard — without needing to remember the optimal prompt.

---quiz question: How does the /ship command demonstrate command composition? options:

• { text: "It runs faster than individual commands", correct: false }
• { text: "It chains /review, /test, /security, and /docs together with quality gates between each step", correct: true }
• { text: "It replaces all other commands", correct: false } feedback: The /ship command composes individual skills into a complete pre-merge workflow with quality gates — if review score is too low, it stops. If tests fail, it stops. This replaces a manual 30-minute checklist with a single command.

---quiz question: Why should teams share skill definitions in a common repository? options:

• { text: "Because skills don't work in individual projects", correct: false }
• { text: "New members get expert-level tools immediately, best practices propagate automatically, and knowledge persists when people leave", correct: true }
• { text: "To reduce the number of API keys needed", correct: false } feedback: Shared skills encode organizational knowledge into reusable AI workflows. New team members are immediately productive with expert-level tools, and the knowledge persists even when the person who created it leaves.