TexTab

The core of TexTab is the action. You write a prompt, name it, and bind it to a keyboard shortcut. There's no limit on how many you create. Keep one for "make this friendlier," another for "expand into bullet points," and a third for "translate to Spanish." Each action fires from any app the moment you press its shortcut.

TexTab doesn't care what's on screen. Select text in a browser, an email client, an IDE, or a plain notes app, then run your action. The output replaces your selection in place, which means no copy-paste round trip and no lost formatting context. If the app has a text field, TexTab can reach it. That's the whole point.

You connect directly to OpenAI, Anthropic, Claude, Groq, Perplexity, or OpenRouter with your own key. There's no TexTab subscription, and no middleman handling your requests. The tradeoff? Costs move to your provider account. Heavy use on an expensive model shows up on that bill instead of a flat monthly fee.

Because your key talks straight to the provider, your text never passes through a TexTab server, which is a real difference from chatbot web apps that quietly log entire conversations on someone else's infrastructure. For people handling sensitive drafts, contracts, or client work, that direct path matters.

A one-click prompt improver rewrites your instruction before it reaches the model, turning something vague like "make this better" into a clearer, more structured prompt that tends to produce noticeably better output. It's a small feature that saves you from learning prompt craft.

Beta plugins add tools beyond text transformation: a chat window, a QR code generator, an image converter, and a color picker. They extend TexTab past pure writing tasks, so the menu bar app doubles as a small utility belt for everyday Mac work.

TexTab ships with ready-made actions so you don't start from a blank page. Fix Grammar, Rephrase, Shorten, Formalize, and Translate are there from the start, and you can edit or duplicate any of them. It's the fastest way to get value before you've written a single custom prompt. No guesswork required.

Connect Perplexity and you can run queries that need live data, with formatted results that include links and images. That turns a text shortcut into a light research tool for facts and current information the base model wouldn't know.