some suggestions

[michabbb]

• add tab completion for folders and subfolders (like a regular linux shell)
  because without that adding files requires either changing directories all the time or diving into a copy&paste hell ;)
• add tab completions for internal commands like /q should show /quit
• STRG+C should exit without any errors/tracebacks
• add /copy and /paste to work with the hosts clipboard

thanks for this project ! 👍

[fry69]

Some more suggestions for commands -> https://aider.chat/docs/usage/commands.html

Also git integration (for /undo, via GitPython) -> https://aider.chat/docs/git.html

And more LLMs than only OpenAI's (via LiteLLM) -> https://aider.chat/docs/llms.html

Maybe even a clever "repository map", which helps the LLM to know about other files in the project? -> https://aider.chat/docs/repomap.html

And there is so much code already existing for inspiration -> https://github.com/paul-gauthier/aider

[kalebzaki4]

Thanks for the feedback and the great suggestions! Let's address each point:

Tab completion for folders and subfolders:

This would be a fantastic improvement! Implementing tab completion similar to a regular Linux shell would streamline file navigation and prevent the need for frequent directory changes or tedious copy-pasting. This is definitely something we’ll consider for improving user experience.
Tab completion for internal commands (/q -> /quit):

Excellent idea! Tab completion for internal commands like /q expanding to /quit would make the interface more intuitive and reduce typing errors.
STRG+C to exit without errors or tracebacks:

Agreed! CTRL+C (or STRG+C in some keyboards) should exit gracefully without throwing errors or displaying unnecessary tracebacks. We'll prioritize making the exit behavior smoother.
/copy and /paste to interact with the host clipboard:

A /copy and /paste command that integrates with the host clipboard would make transferring data more efficient. It’s a useful suggestion for anyone who frequently works between different tools or documents.
Thanks again for all these ideas! We’ll explore the possibility of integrating Git functionality (e.g., for /undo via GitPython), support for more LLMs beyond OpenAI (using LiteLLM), and the idea of a "repository map" that helps the model understand the project structure better. Plus, we appreciate the links and existing code you’ve pointed to for further inspiration.

If you have more thoughts, feel free to share. Thanks for supporting the project! 👍