## Definitely

* Switch to "process" provider for AWS credentials (much less hacky)
* Session timeout (plain duration, or activity-based?)
* Fix rehide behavior when new request comes in while old one is still being resolved
* Additional hotkey configuration (approve/deny at the very least)
* Logging
* Icon
* SSH key handling

## Maybe

* Flatten error type hierarchy
* Rehide after terminal launch from locked
    * Generalize Request across both credentials and terminal launch?
* Make hotkey configuration a little more tolerant of slight mistiming