[Bug]: F looks for new suggestion after fixing the previous command

[swirle13]

What happened?

Given export _PR_MODE="noconfirm" is set in ~/.zshrc before eval "$(pay-respects zsh)", running a bad command, git sattus and fixing it via f causes it to fix and run the correct command, but afterwards, outputs the "looking for new suggestion..." message for the fixed command.

A separate but related issue (which I can make a separate report for) will occur when running without export _PR_MODE="noconfirm" where the main thread panics due to InvalidCertificate(UnknownIssuer)

I'm not sure why it needs a certificate. I'm also not sure why it doesn't throw this issue when setting to "noconfirm" mode. Is there an internet request to find more suggestions when using the standard mode? I've ran into certificate issues before due to my work VPN and I've passed the exported cert to VSCode and other tools that need it. If necessary, is there a path to do this for pay-respects?

The command and output

❯ git sattus
git: 'sattus' is not a git command. See 'git --help'.

The most similar command is
        status

~ 
❯ f
git status
fatal: not a git repository (or any of the parent directories): .git

Looking for new suggestion...

No suggestion found for command: git status

If you think there should be a suggestion, please open an issue or send a pull request!
https://github.com/iffse/pay-respects

Build information

pay-respects: v0.7.8

macOS 15.5

ZSH 5.9

No similar issue

• I have searched the issue list and there is no similar issue found

[iffse]

The "fixed command" didn't run successfully, and that's why it's looking for new suggestions. If you run the same case inside a git repository such that git status exit successfully, you won't get this message.

With no confirm, AI module is skipped as the suggestions can be anything from legit to garbages.

The internet connection is for getting suggestions from LLM models by the AI module. Haven't looked up yet how to use a custom certificate.

[swirle13]

The "fixed command" didn't run successfully, and that's why it's looking for new suggestions. If you run the same case inside a git repository such that git status exit successfully, you won't get this message.

Ah okay, so it's triggering that logic based on the previous command's exit code? Thank you for the explanation.

The internet connection is for getting suggestions from LLM models by the AI module. Haven't looked up yet how to use a custom certificate.

I'm sure I could look into the source code for the answer, but what LLM modules are being used? Do you have any data privacy statements or controls in place? I know for myself, and possibly others, I'm using this in a professional environment and I'd like to know if I can control or restrict the data being sent from my computer to an unknown LLM. I'm sure it's fine and I don't see any reason you'd be malicious, but companies are really careful about their data, so I don't want to be violating any policy since this uses an LLM via a request over the internet.

[iffse]

Current LLM provider I am using is GroqCloud which does not retain your data. You can always use your own LLM provider or local models like Ollama so it never goes out of your computer