slash command
A short command starting with a slash, with which you issue a pre-prepared, longer instruction or built-in function to an AI agent or chatbot with a single word.
In plain terms
A slash command is like a regular customer's order at a coffee shop: it is enough to say 'the usual', and the barista knows this means a long coffee with oat milk, no sugar. Likewise, with an agent it is enough to type /summary, and it sees a pre-written, detailed instruction behind the word.
It is useful so that frequently repeated tasks do not need to be re-explained every time. You write the long request once, or someone in the team writes it, and from then on everyone calls it up with the same short command.
It differs from casual chat in that the slash signals to the program: this is not a question or conversation, but a named command. Some commands are built-in, such as deleting the conversation or opening settings, while others you can create yourself.
How it works
- You type a slash into the text field, and the program usually offers a list of available commands.
- You select or type the command name, and if needed, add details afterward, such as a filename or a topic.
- The program searches for what belongs to the name: either a built-in feature or a saved instruction text, often stored in a simple text file.
- When it comes to instruction text, the program inserts it into the conversation along with your additions, as if you had typed the whole thing.
- The agent then executes the task just like any other request: it reads, uses tools, and provides an answer.
An everyday example
In an office, colleagues ask the AI agent for a weekly report every Monday. Instead of everyone describing what they want in their own words, they type: /weekly-report, and the agent always prepares it in the same structure with the same criteria.
Who is affected
A common misconception
Limits and risks
- Commands vary by program: what works in one tool may be unknown in another or have a different meaning.
- A command is only as good as the instruction behind it; an inaccurately written template produces the same error every time it is used.
- If you run a command created by someone else, you may not know what instructions it gives the agent, so it is worth reading its content before use.
- The agent can make mistakes even after your command; you must check its result just as you would any other answer from it.
Similar, but not the same
- prompt – A prompt is any freely formulated request, while a slash command is a short, name-callable form of a saved prompt or function.
- agent skill – The agent can invoke the skill itself if it fits the task, while the slash command is typically initiated by you explicitly. In some tools, the skill can also be triggered with a slash, so the boundary between the two concepts blurs there.
- keyboard shortcut – A keyboard shortcut is pressing keys together; a slash command is typed text to which you can add further details.
Related concepts
AI-agent, user interface
custom command, built-in command
prompt, promptsablon, agent skill, coding agent, system prompt, command line
A brief history
Slash-prefixed commands were used by earlier chat systems, but the IRC, created by Jarkko Oikarinen in 1988, made them widely known. For example, the /join command allowed users to enter a channel. Later, workplace and community chat programs such as Slack and Discord adopted them, followed by AI assistants and coding agents.
Where things stand
As of late 2026, this is a standard, basic way of interacting with most coding agents and workplace chat programs. Custom commands are increasingly merging with agents' skills, that is, reusable, shareable instruction packages.
In the news
Also known as
slash command · slash parancs · perjeles parancs · perjelparancs · /parancs · slash commands · custom command · custom command
Sources
This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.