shell_gpt

So this is actual sorting (i.e. the same amount of lines appear in the output)? Otherwise it might be more or less the same as using https://github.com/TheR1D/shell_gpt ?

I use a cli tool 100x/day for misc things; can hit openai or a local llm. https://github.com/TheR1D/shell_gpt Has nice shell integration.

I was wondering if anyone else is using a minimal terminal LLM. What I mean is a small cli application that lets you use stdin/stdout/etc to just pipe data to/from an llm (provider agnostically). For me this method has become my favorite way of using of LLMs (when I only need text IO) - but I suspect that I've (once again) just reinvented the wheel here. Previously I used https://github.com/TheR1D/shell_gpt, but I ran into a few reliability and usability issues with it. The way this works for me, is that I just launch a terminal which automatically starts a new LLM session, and I can use the CLI to load and manipulate previously saved sessions. Maybe in the future, what I see is that one could launch a generic LLM daemon on one's computer to just expose an API for any other tool to use, like an equivalent of an LSP, but for LLM instead of

(aichat in tangential comment looks much better) Besides aider and ollama, I think shell_gpt https://github.com/TheR1D/shell_gpt is great for quick chats / lookups. Being able to quickly cat files to repl sessions saves a lot of time. I need to integrate distil whisper large v3, aider, and shell_gpt to tidy up a lot of my disjointed LLM use. As someone else mentioned, the commits created by aider allow me to "skip" some intermediate steps that would be required when working on coding tasks with other frameworks.

I've been using shell-gpt[1] for the same, and it is almost irritatingly useful. I fear that my somewhat decent shell-fu is going to atrophy pretty rapidly in this new world. in my experience, this is the kind of thing that LLMs are great for - small, one-off tasks with clearly defined parameters. (and, with careful application - low stakes.) [1]: https://github.com/TheR1D/shell_gpt

Right, but it's certainly easier for people who might not even know what "API" stands for, and that's quite nifty. As far as self-hosted frontends go, I can personally recommend SillyTavern[1] in the browser, ChatterUI[2] on mobile, and ShellGPT[3] for CLI. LobeChat looks pretty cool, though! I'll definitely check it out. [1] https://github.com/SillyTavern/SillyTavern [2] https://github.com/Vali-98/ChatterUI [3] https://github.com/TheR1D/shell_gpt

Nice! How does it compare with https://github.com/TheR1D/shell_gpt ?

I'm also a very heavy daily user of a similar tool, shell_gpt. Not having it would feel like a limb missing, very odd. https://github.com/TheR1D/shell_gpt

How is this different from ShellGPT - https://github.com/TheR1D/shell_gpt ?

I use a random one [1], when I remember. Type what I need in english and press Ctrl-G: ~/ git clean branch⌛ (ctrl-g gives me the little hourglass) wait for a second and the command line is replaced: ~/ git clean -f -d && git checkout branch Of course when the result is not what you want, might as well open the browser... [1] https://github.com/TheR1D/shell_gpt

For people who use terminal a lot, I find shell gpt a good mixture [1]. Flexible enough to pipe commands, generic enough to ask questions. [1] https://github.com/TheR1D/shell_gpt

https://github.com/TheR1D/shell_gpt already exists

ShellGPT https://github.com/TheR1D/shell_gpt does pretty well for a lot of use cases. I mostly use it in REPL mode, switching topics as needed. I have wrappers around the `sgpt` command to, say, start a REPL with a particular topic, say, Python, which loads my previous history on that topic as part of the prompt. I also have an alias to save existing chats as text files so I can go back and review history. Finally, there is an alias to load a question up in an editor if I need to enter multiline text, e.g. to discuss code fragments, etc. I expect command-line workflows to be pretty individualized and I'm curious what others do. For me (old programmer), using a command line REPL feels much more natural (and blissfully noise-free) than going to a Web page to talk to, say, ChatGPT.