Fair question for any "AI tool": where do my text and my key actually go? With Prompt Enhancer there's no backend of mine in the loop. You bring your own Claude API key, it lives in your OS keychain, and your selected text goes straight from your machine to Anthropic. I never see either one. I keep going back and forth on whether to also offer a hosted version so I'm curious: is bring-your-own-key a dealbreaker for you, or the whole reason you'd trust it?

Quick feature I'd love feedback on: the preset system.

Right now there are four Default, Concise, Verbose, and Code so the same hotkey restructures your text differently depending on what you're doing. Writing a commit message vs. drafting a spec need very different prompt shapes.

Why I built this:

I kept firing off lazy one-liners like "fix the auth bug" and then spending minutes re-explaining what I actually meant. The problem was never the model it was my prompt structure. But manually writing role / context / task / output every time, in every app, is tedious enough that nobody does it.

One thing I'm proud of: Prompt Enhancer works anywhere you type Claude, ChatGPT, Cursor, Notion, Figma, your terminal, an email draft. No integrations to set up, no copy-pasting into a separate web app.

It's a native desktop tool (Swift on macOS, Tauri/Rust on Windows), so the hotkey just works system-wide. Highlight, press, done.

The pricing question always comes up, so let me be upfront: Prompt Enhancer is free forever if you bring your own Claude API key.

No subscription, no per-prompt markup, no account wall. Your key lives in your OS keychain, and your text goes straight to Anthropic nothing routes through our servers. You pay Anthropic directly for what you actually use (usually fractions of a cent per prompt).