AgentsKit
PromptsBlogUse casesFree toolsGet AgentsKit
Home/Blog/Slash commands
Copy-paste examples

15 Claude Code Slash Command Examples

July 202610 min read15 working commands included

A custom slash command is the simplest extension Claude Code has: a markdown file in .claude/commands/ whose filename becomes the command and whose body becomes the prompt. No registration, no build step, save the file, type / and it's there. What most people are missing isn't syntax, it's a starting library. Here are fifteen commands you can paste today, organized by the work they do.

The 60-second syntax primer

.claude/commands/fix-issue.md
---
description: Find and fix a GitHub issue by number
argument-hint: <issue number>
---

Fix GitHub issue #$ARGUMENTS.
1. Read the issue with `gh issue view $ARGUMENTS`
2. Locate the relevant code and reproduce the problem
3. Implement a fix, add a regression test, run the suite
4. Commit with message "fix: <summary> (#$ARGUMENTS)"

$ARGUMENTS captures everything typed after the command (/fix-issue 421);$1, $2 grab positional arguments individually. Lines starting with ! run shell commands and inject their output into the prompt, and @path/to/file pulls a file in as context. The frontmatter is optional but the description shows up in the command menu, so write it.

Code quality (1–4)

1. /review, "Review the current diff (!git diff HEAD) for correctness bugs first, style second. For each finding: file, line, what breaks, and a concrete fix. Rank by severity. If nothing is wrong, say so, do not invent findings."

2. /refactor, "Refactor $ARGUMENTS without changing behavior. Tests must stay green, run them before and after. List each change and why it improves the code. If a change would alter behavior, stop and report instead."

3. /test, "Write tests for $ARGUMENTS following this repo's existing test patterns. Cover the happy path, edge cases, and one failure mode. Run them and fix failures before finishing."

4. /explain, "Explain how $ARGUMENTS works: entry points, data flow, key invariants, and anything surprising. Audience: a developer new to this codebase. No code changes."

Git and shipping (5–8)

5. /commit, "Stage and commit the current changes. Group unrelated changes into separate commits. Messages follow conventional commits; the subject line explains why, not just what."

6. /pr, "Create a pull request for this branch: summarize the diff against main, list notable decisions and trade-offs, add a test plan, and open it with gh pr create."

7. /changelog, "Update CHANGELOG.md from commits since the last release tag (!git log $(git describe --tags --abbrev=0)..HEAD --oneline). Group by Added/Changed/Fixed. Write for users, not committers."

8. /release-check, "Pre-release audit: run tests, typecheck, and build; check for TODOs and console.logs in the diff; verify version bumps are consistent. Output a go/no-go verdict with a checklist."

Debugging and maintenance (9–12)

9. /debug, "Debug this: $ARGUMENTS. Reproduce first. State your hypothesis before changing anything, add instrumentation to confirm it, then fix the root cause, not the symptom, and prove the fix with a test."

10. /deps, "Audit dependencies: flag outdated packages, known vulnerabilities (!npm audit), and anything unmaintained. Recommend upgrades in order of risk, and note breaking changes from changelogs."

11. /todo-sweep, "Find every TODO/FIXME/HACK in src/. For each: still relevant? If trivially fixable, fix it. Otherwise output a prioritized list with file and line references."

12. /dead-code, "Find unused exports, unreachable branches, and orphaned files. Verify each candidate is truly unused (check dynamic imports and string references) before listing it for deletion."

Documentation and beyond (13–15)

13. /docs, "Update documentation for $ARGUMENTS: check README and docs/ against the current behavior of the code, fix drift, and flag anything undocumented that a new user would need."

14. /onboard, "Give me a tour of this codebase: purpose, stack, the five most important files, how data flows through a typical request, and what I should read first. Then suggest three good first tasks."

15. /standup, "Summarize what changed in this repo in the last 24 hours (!git log --since=yesterday --oneline): what shipped, what's in progress, anything blocked. Three bullets, written for a standup."

Patterns that separate good commands from noise

  • Encode the checklist, not just the ask. "Review my code" is a wish. Numbered steps with an output format is a workflow that runs the same way every time, that consistency is the entire value of a command.
  • Inject fresh context with !. A command that pulls the actual diff, log, or audit output grounds Claude in reality instead of memory.
  • Define "done". End with the verification step, run the tests, show the output, produce the verdict. Commands without a finish line produce plausible-looking half-work.
  • Commit them to the repo. Project-level commands are shared team workflows in git. Your best reviewer's checklist, everyone's /review.

These 15 are the free sample: the AgentsKit kits ship 181 slash commands covering review, testing, security, refactoring, docs, marketing, and more, installed in one command. Browse the full library →

FAQ

Slash command or skill, which should I make? +
Commands run when you type them; skills fire automatically when the task matches. If the workflow is deliberate ("now do a release check"), make it a command. If forgetting to invoke it would hurt ("always follow our migration rules"), make it a skill.
Can commands take multiple arguments? +
Yes, $1, $2, etc. for positional arguments, or $ARGUMENTS for the whole string. Add an argument-hint in frontmatter so the menu shows what to pass.
Why isn't my command showing up? +
Check the location (.claude/commands/ in the project or ~/.claude/commands/ globally), the extension (.md), and restart the session. Subdirectories create namespaces, commands/git/pr.md appears as a project command with "git" in its description.

Keep reading

Skills vs Subagents vs Slash Commands vs MCP

When a slash command is the right primitive, and when it isn't.

Read →

CLAUDE.md Best Practices

The exact template to give every new project.

Read →

The 12 Subagents Worth Setting Up First

Commands trigger workflows; these agents do the heavy lifting inside them.

Read →
AgentsKit

Your AI engineering & marketing team for Claude Code.

Product

What's insideKitsPricingFAQ

Resources

Claude promptsBlogUse casesComparisonsFree tools

Legal

LegalTermsPrivacyRefundsDisclaimer

Connect

epictools.io@gmail.comagentskit.coX (Twitter)
Unofficial & independent. AgentsKit is not affiliated with, endorsed by, or sponsored by Anthropic. "Claude," "Claude Code," and "Anthropic" are trademarks of Anthropic. Content on this site is for educational purposes, see our disclaimer.

© 2026 AgentsKit. All rights reserved.