Home > Features > Slash-Command Prompts
If you've ever typed / in Slack or Notion and watched a menu pop up, you already understand the mechanic this page is about — just aimed at a different problem. Instead of inserting an emoji or a table, typing // in ChatGPT, Claude or Gemini opens a small palette of your own saved prompts and drops the one you pick straight into the chat box. No copy-paste, no tab switching, no retyping the same instructions you've already written a hundred times.
This is what "slash-command prompts" means in practice, why it beats the tools people currently duct-tape together for the same job, and exactly how it works across the ten AI sites PromptDock supports.
A slash command is a short trigger — usually a symbol plus a letter or two — that a piece of software watches for while you type, and reacts to instantly. Discord uses / for bot commands. Notion uses / to insert blocks. Superhuman uses keyboard shortcuts for canned email replies. The underlying idea is always the same: stop making the user navigate a menu, and let a trigger they can type without thinking bring the menu to them.
Applied to AI chat, a slash-command prompt is a saved piece of text — a full prompt, an instruction, a template — that you can summon and insert by typing a trigger inside the message box itself. PromptDock uses // specifically because it's a sequence nobody types by accident in normal conversation with an AI, so it never fires when you don't want it to. The moment you type the second /, a small dropdown appears right above your cursor, filtered live as you keep typing.
The result is that your best prompts stop being something you remember, search for, or keep in a separate document. They become something you invoke — the same way a keyboard shortcut is faster than clicking through a menu, once your fingers learn it.
Plenty of people try to solve this with tools that weren't built for it, and the seams show up fast:
contenteditable editors, which is exactly where these tools are most likely to misbehave — double insertions, stripped line breaks, or a macro that just doesn't fire because the editor swallowed the keystroke differently than a plain <textarea> would.The whole interaction is designed to take about two seconds, with no mouse required:
// characters, no manual cleanup.Pinned prompts float to the top of the unfiltered list, so the two or three prompts you use every single day are the first thing you see even before you type a filter letter. Folders and tags organize the rest, so a 60-prompt library doesn't turn into a scroll-fest.
A saved prompt that never changes is useful. A saved prompt with blanks you fill in each time is far more useful, because it turns one entry in your library into an unlimited number of variations. PromptDock supports this with a simple {{variable}} syntax:
| Saved template | What happens on insert |
|---|---|
Write a LinkedIn post about {{topic}} for {{audience}} | PromptDock prompts you for topic and audience as it inserts, then drops in the completed text. |
Summarize this in {{number}} bullet points, tone: {{tone}} | Fill in "5" and "casual" once, reuse the same template for every summary you ever need again. |
Act as a {{role}} and review this for {{focus_area}} | One code-review prompt becomes a security review, a performance review, or a style review just by changing two words. |
This is the piece a plain text expander can't really replicate without a lot of manual macro-scripting — and it's the difference between "I have a folder of prompts" and "I have a small library of tools that adapt to whatever I'm doing today."
The single biggest limitation of most prompt tools is that they're built for one site. PromptDock's // palette and your saved library are the same across all ten of the platforms it supports:
Save a prompt once — from any of the ten, or by right-clicking selected text on any page and choosing "Save to PromptDock" — and it's available with // in all of them. If you draft in ChatGPT, fact-check in Perplexity, and do a final pass in Claude, your prompt library follows you across that whole workflow instead of living in just one of those tabs. For a deeper look at why keeping prompts trapped in a single chat history is such a common pain point, see reusing prompts across ChatGPT, Claude and Gemini.
Typing // is the fast path, but it's not the only path. PromptDock also opens the exact same palette with a configurable keyboard shortcut, which matters in a couple of real situations: if you're deep in a sentence that happens to contain a double slash for some other reason, or if you simply prefer a modifier-key habit like the command palettes in VS Code or Raycast. Either way you land in the identical searchable list — the trigger is just a preference, not a different feature.
Here's the honest side-by-side, because "slash commands" as a concept is not unique to PromptDock — what differs is whether the tool was actually built for AI chat boxes or adapted to them after the fact:
| PromptDock | Generic text expander | Notes app + copy-paste | |
|---|---|---|---|
| Built for AI chat editors | Yes — purpose-built for ChatGPT/Claude/Gemini's rich editors | No — built for ordinary web forms, glitches in contenteditable boxes | N/A, manual |
| Fill-in-the-blank templates | Yes, native {{variables}} | Only with manual macro scripting | Manual editing every time |
| Works across 10 AI sites, one library | Yes | Depends on the site's editor | Yes, via copy-paste |
| Account required | Never | Usually, for cloud sync | Depends on the app |
| Where data lives | 100% local on your device | Usually the vendor's cloud | Depends on the app |
| Price | Free tier, $19 once for Pro | $3–8/month, ongoing | Free (if you already use the app) |
Not for personal reusable prompts. ChatGPT has no native // or /command system for inserting your own saved text — that functionality comes from an extension like PromptDock, which adds the // trigger directly inside the chat box.
It's a real command palette, not a text-replacement macro. PromptDock's content script watches the chat input, detects // as you type, and renders a searchable dropdown of your saved prompts right there — nothing is sent anywhere, and the palette disappears once you pick or cancel.
Generic text expanders are built for ordinary web forms, and AI chat boxes are rich contenteditable editors that commonly break their expansion logic. PromptDock's palette is purpose-built for these specific editors, so insertion is reliable, and it also supports fill-in-the-blank templates and folders that expanders don't.
Yes. If // ever conflicts with something you type naturally, PromptDock also opens the same palette with a configurable keyboard shortcut, so you're never stuck with only one way in.
The same // palette and the same saved library work in ChatGPT, Claude, Gemini, Perplexity, DeepSeek, Grok, Microsoft Copilot, Mistral, Poe and Google AI Studio — ten sites total, all from one extension install. See the full breakdown in how PromptDock compares to AIPRM and other ChatGPT-only tools.