How to Add Your Own LLM to aish: A Practical Guide from Selectel

⚡ The gist in 5 seconds - Selectel has released a practical guide on connecting your own LLM to aish — an AI agent that lives in the terminal of a server OS. - The article was published on Habr by project manager Natasha and continues a previous piece about why this agent was built in the first place. - Specific steps, requirements, and limitations are in the article itself; the link is at the end of this post. ### 🔍 What was found We're talking about aish — an internal Selectel project, an AI agent that lives in the terminal of a server OS. The first article explained the motivation: why the company needed such a tool. The new text is no longer about the idea but about practice: how to hook your own model up to the agent. The format is typical for Selectel's corporate blog on Habr: the author writes in the first person, without marketing fluff, and walks the reader through the setup step by step. What exactly needs to go into the config files, which models are supported, and what pitfalls exist — all of that is left out of the announcement; the details are only in the full article via the link. ### 💡 Why it matters Terminal AI agents are no longer an experiment: they are already a convenient way to handle server routine — from log analysis to command generation. The key question for any infrastructure team is which model powers all of this and where the data goes. The ability to plug in your own LLM removes exactly this blocker: you can keep traffic inside your own perimeter or on your own API key. How smoothly the connection works in practice — Selectel shows in the guide.