[feat] Direct file output for LLM responses

[PeterDaveHello]

Hi,

First, this project is cool! The command-line interface for LLMs is handy.

I wonder if you would consider adding a feature to write LLM output directly into a file? This could be very helpful for users who want to save responses for later reference, processing, or as a final result of their interaction with the LLM.

A possible implementation could be a command-line option like -o filename.txt or --output filename.txt.

Thank you for your great work and for considering this suggestion.

[simonmysun]

Thank you for your appreciation!

If you want to save to a file, you can redirect stdout or pipe the output to tee, depending on you want to see the result in your terminal or not.

ell "What is the capital of France" | tee result.txt

or if you simply don't need the result to be shown in your terminal:

ell "What is the capital of France" > result.txt

But adding another option for output is also a good idea. This will allow user to have more flexibility. Thank you for the suggestion!

[PeterDaveHello]

Yeah, it's not too complex to do the pipe things manually 😆 but an option would also be great 😄

[simonmysun]

Question: What's the expected behaviour in record and interactive mode?

[PeterDaveHello]

Thanks for considering the feature! How about we record both the input and output of the LLM in both record and interactive modes? What do you think?

[simonmysun]

Implemented.