LLM Chat via SSH

[demosthanos]

Skimming the source code I got really confused to see TSX files. I'd never seen Ink (React for CLIs) before, and I like it!

Previously discussions of Ink:

July 2017 (129 points, 42 comments): https://news.ycombinator.com/item?id=14831961

May 2023 (588 points, 178 comments): https://news.ycombinator.com/item?id=35863837

Nov 2024 (164 points, 106 comments): https://news.ycombinator.com/item?id=42016639

[ccbikai]

Many CLI applications are now using Ink to write their UIs.

I suspect React will eventually standardize all UI writing approaches.

[amelius]

I'd rather apt-get install something.

But that seems not a possibility in the modern days of software distribution, especially with GPU-dependent stuff like LLMs.

So yeah, I get why this exists.

[halJordan]

What is the complaint here? There are plenty of binaries you can invoke through your cli that will query a remote llm api

[gsibble]

We made this a while ago on the web:

https://terminal.odai.chat

[gbacon]

Wow, that produced a flashback to using TinyFugue in the 90s.

https://tinyfugue.sourceforge.net/

https://en.wikipedia.org/wiki/List_of_MUD_clients

[dncornholio]

Using React to render a CLI tool is something. I'm not sure how I feel about that. It feels like like 90% of the code is handling issues with rendering.

[demosthanos]

I mean, it's a thin wrapper around LLM APIs, so it's not surprising that most of the code is rendering. I'm not sure what you're referring to by "handling issues with rendering", though—it looks like a pretty bog standard React app. Am I missing something?

[xigoi]

It’s not clear from the README what providers it uses and why it needs your GitHub username.

[ccbikai]

Connects to any OpenAI-compatible API.

Using a GitHub username prevents abuse.

[gclawes]

Is this doing local inference? If so, what inference engine is it using?

[demosthanos]

No, it's a thin wrapper around an API, probably OpenRouter or similar:

https://github.com/ccbikai/ssh-ai-chat/blob/master/src/ai/in...

[ccbikai]

Currently using the OpenAI API to access multiple models. You can use ollama to access local inference models

[ryancnelson]

this is neat.... whose anthropic credits am i using, though? sonnet-4 isn't cheap! would i hit a rate-limit if i used this for daily work?

[ccbikai]

I am the author, thank you for your support.

Welcome to help maintain it with me

[kimjune01]

hey i just tried it. it's cool! i wish it was more self aware

[ccbikai]

Thank you for your feedback; I will optimize the prompt

[t0ny1]

does this project request to llm providers?

[cap11235]

Are you serious? Yeah, its using gemini 2.5 pro without a server, sure yeah.

[eisbaw]

Why not telnet?

[accrual]

I'd love to see an LLM outputting over a Teletype. Just tschtschtschtsch as it hammers away the paper feed.

[RALaBarge]

No HTTPS support