Commands & Slash Integration

Skills and slash commands have merged. When you create a skill, it automatically registers as a slash command. If your skill is named "deploy-staging," you can invoke it by typing /deploy-staging in the Claude Code terminal. This is more than a convenience — it changes how you think about skills. Instead of relying on Claude to pattern-match your trigger phrases against skill descriptions, you can invoke any skill explicitly and instantly. The slash command is the reliable path. The trigger phrase is the natural-language fallback.

Building a command library is the infrastructure equivalent of building a toolbox. Start with the workflows you repeat daily: deploying, testing, committing, auditing. Create a skill for each one. Then build higher-order commands that compose the basics: /full-deploy might invoke /test first, then /deploy-staging, then /smoke-test, then /deploy-production. Each skill is a building block. Each composition is a new capability. Over time, your command library becomes a custom development platform tailored to your project.

1. Step 1: Identify Repeated Workflows Track what you do more than twice a week. Deploy, test, commit, format, audit, scaffold — these are all skill candidates.
2. Step 2: Create Atomic Skills Each skill does one thing well. /test runs the test suite. /lint runs the linter. /check runs the type checker. Keep them focused.
3. Step 3: Compose Higher-Order Commands Build commands that chain atomic skills. /ship runs /check, /lint, /test, and if all pass, commits and pushes. The composition is itself a skill.
4. Step 4: Share with the Team Skills live in .claude/commands/ and can be committed to version control. Every team member gets the same command library on clone.