Support evil movement

[jasonatwell]

Hi, thanks for the project, really excited about it! I use evil-mode and have noticed an issue. To reproduce

1. turn on evil-mode
2. enter insert mode and enter a few character
3. enter normal mode and move your cursor back 1 character
4. type one character
5. expect that one character to be inserted at cursor, but instead it is inserted at the end of the string you typed in step 2 above

Here's a demo in case visuals help

Screen.Recording.2026-04-06.at.15.44.15.mov

[jasonatwell]

I'm not sure how related this is so if you want me to post a new issue I can, but I can delete all text in the terminal while in insert mode using ctrl+w, see demo

Screen.Recording.2026-04-06.at.15.46.10.mov

[michaelBelsanti]

Something similar to evil's vterm module would be needed. I was just trying to write a module for hel, which also has its own vterm module that was probably based on evil's, so it has the same needs as evil.

I think there's a few functions that would be needed for evil support that's equivalent to vterm:

1. ghostel-goto-char - takes a buffer position and moves the shell cursor to that position
2. ghostel--get-prompt-point - returns buffer position of the start of editable input
3. ghostel--get-end-of-line - returns buffer position of the end of editable input
4. ghostel-cursor-in-command-buffer-p - predicate to check if cursor is in editable region
5. ghostel-delete-region - pty aware region deletion to avoid deleting read-only text
6. ghostel-goto-cursor - snap emacs point to the terminals cursor pos

its possible I missed something and some of these can be approximated with existing ghostel functions

[dakra]

Thanks for the pointers @michaelBelsanti

Can someone try out #54 and give feedback?
(NOTE: You need to manually compile the native module first since I didn't release an artifact with the latest ghostel--cursor-position change yet.)

I'm not a evil user but that change looks small enough that it doesn't seem to cause maintenance burden.