Slash Commands
What it is
A slash command is a reusable prompt you trigger by typing /name in your agent. Instead of retyping a long, carefully worded instruction every time, you save it once and invoke it with a short name. Many commands accept arguments, so a single command can be reused with different inputs — /review src/auth versus /review src/api.
Strengths
- Saves a good prompt once and replays it reliably, instead of rewriting it each time.
- Short to invoke, so common workflows become a few keystrokes.
- Arguments make one command flexible across many inputs.
- Stored as plain files you can version and share, so a team uses the same wording.
Trade-offs
- A command is only as good as the prompt inside it — a vague one gives vague results.
- A growing pile of commands gets hard to remember and maintain.
- It's a shortcut to a prompt, not a guarantee; the agent still interprets it.
- Slightly hides what's actually being sent, which can surprise newcomers.
When to use it
Make a slash command whenever you find yourself typing the same instruction repeatedly — a code review prompt, a commit-message generator, a "explain this file" request — especially one with a fixed structure you want to stay consistent.
Vibe coding fit
Slash commands turn your best prompts into a personal toolkit. The careful prompt you wrote once becomes a reliable button you press, so quality doesn't depend on remembering the exact wording each session. Pair them with arguments to cover whole families of tasks with one well-tuned command.
---
description: Review a file for bugs and clarity.
argument-hint: <path>
---
Review the file at $ARGUMENTS. List concrete bugs first,
then smaller clarity or naming suggestions. Be specific.