what is atuin-ai-server fr?

Atuin AI Server is a self hosted backend for Atuin's AI chat feature, which normally runs as a hosted service the Atuin team operates. This project lets someone run that same chat engine themselves, pointed at any AI model server that speaks the common OpenAI style chat format, including popular local options like Ollama, LM Studio, and llama.cpp, or a compatible cloud API. Running it is meant to be simple. It ships as a Docker image, so a user copies an example configuration file, edits it to point at their chosen model server, and starts the container with one command. It can also be built and run directly from source code for people who prefer not to use Docker, using the Elixir and Erlang programming languages and the Gleam tool that the project is built with. Configuration lives in a single file listing the server's port, the address of the model backend, and one or more named models the Atuin command line tool can choose from. Whatever model is chosen must support what is called tool calling, meaning the model can request that certain actions be performed on its behalf, since Atuin's chat feature relies on this at every turn. The server also supports a few optional built in tools of its own, letting the AI search the web or scrape a webpage, each of which only turns on if the matching service key is provided in the configuration. Once running, a user points their local Atuin configuration file at this server's address, and their command line chat sessions run against the model server they chose instead of Atuin's hosted default. The project intentionally keeps no accounts, no stored history, and no usage limits, since it is meant purely as a private bridge between the Atuin tool and whatever model infrastructure someone already has. It is released under the Apache 2.0 license, meaning it can be freely used, modified, and redistributed as long as the license notice is kept.