Slash commands¶

Slash commands allow you to control coding agents quickly and efficiently using the keyboard. Type / in the editor to open the slash pop-up, select a command, and your coding agent will carry out actions such as switching models, changing permissions, clearing the context or executing a specific workflow, without you having to leave the input window.

Here are the links to the documentation for slash commands for three popular coding agents:

Skills vs. Slash Commands¶

At first glance, slash commands and skills may seem similar. However, they operate on completely different levels and address different issues. The choice between them affects how much manual work you have to do in each session.

The most obvious difference is that skills are triggered automatically depending on what you’re currently doing, whereas slash commands must first be entered manually.

Skills are process-oriented¶

They define the following:

What triggers the skill?
What steps should be followed once the skill is triggered?
What output or action should occur?

The first step is crucial: your coding agent analyses the context of your current activity – which files are open, what is currently being written, what the task description is – and matches this information against the available skills. If there is a match, the skill is executed. This is precisely what makes skills ideal for recurring, predictable tasks that need to be carried out consistently without receiving a prompt every time.

Slash commands are explicit instructions¶

You type /COMMAND, and your coding agent carries out that command. Nothing is executed that you haven’t explicitly instructed it to do.

For example, in Claude, you can use /simplify and /batch to effectively tackle over-engineering in your code. And if you’re struggling with an overloaded context window, you can simply use /compact to prevent context rot.

Custom slash commands¶

You can also create custom slash commands for long, frequently used prompts. This allows you to enter just a short slash command instead of long instructions. To do this, simply type /NEW_COMMAND in Claude Code. Custom slash commands for Claude Code are stored in the .claude/commands/ directory. Each file contains the text of the command. When you enter the slash command, Claude reads the file and executes it. It is at this point that the line between commands and skills begins to blur.

When should you use slash commands?¶

Use a slash command if the task is context-dependent, reactive or one-off. Here are a few situations where slash commands are the right choice, as they should not be triggered automatically but decided by you:

Reducing the context: If a session has been running for a while, the context window is getting longer and longer, and delays or inconsistencies are occurring, you can use /compact in Claude Code to shorten the window.
Temporarily ignoring the context: /btw is intended for situations where you want to ask Claude Code a question in the middle of a task without interrupting the workflow.
Debugging a specific output: If you unexpectedly receive an error message, you can use /simplify just for that section.
Routine tasks: If you want to check your entire codebase for specific patterns once a month using your coding agent, this is a case for a slash command, not a skill.