jmburke1/promptsanitizer ? reverse-engineered prompt

Build me a small local desktop app that helps me sanitize prompts before I paste them into ChatGPT or Claude.

I want a simple Java app with a clean Swing interface where I can keep my own list of sensitive words and phrases, like API keys, email addresses, project names, or anything else I do not want sent to an LLM. The app should let me paste a raw prompt on one side, click a button to replace those values with safe placeholders, then copy the cleaned version out. It should also let me paste the model’s reply back in and restore the original values locally.

Please save my dictionary between sessions in a local JSON file, and make it easy to add, edit, remove, and reorder entries. If it makes sense, include support for regex style replacements too. Keep everything fully offline and local, with no cloud calls or browser extension stuff. If you need to look up current Java or Swing docs online, feel free to do that.