Loading…
Loading…
yo is a personal AI assistant for the terminal, written in Rust. Ask with yo <question>, or pipe command output into a question. It keeps one chat per terminal session and stores chats and durable memory locally in SQLite.
New responses use one gateway chosen during setup: Vercel AI Gateway, LLM Gateway or OpenRouter. The current code has no direct provider integration and no local-model backend. Cached chat and memory are local; generation still needs network access, a valid gateway key and available credits.
The streaming parser handles events across arbitrary response chunks and requires the data: [DONE] completion event. If the connection ends before that event, it returns a stream error instead of treating a partial response as complete. Rate-limit errors preserve retry information when the service provides it.
Command execution uses an OS sandbox on macOS through Seatbelt and on Linux through Bubblewrap, with a separate approval policy selected by the user. Windows binaries are available; that does not imply that these two operating-system sandbox mechanisms apply on Windows.
The source includes Rust-native workflow evaluations, yo doctor diagnostics and signed self-update support. These are implemented checks and update paths. Their presence does not prove that an arbitrary provider request, environment or command will succeed.
Install with Cargo:
Or use the Homebrew tap:
Then choose a gateway:
The repository documents the shell hook for each supported shell.
Ask a question:
To explain a failure, pipe a small non-sensitive log. Replace some-command with the command you are testing:
Use a small non-sensitive fixture to test authentication, rate limits and a dropped stream before relying on a long session.
cargo install yobrew tap Montekkundan/yo https://github.com/Montekkundan/yo.git
brew install Montekkundan/yo/yo-binyo setupyo what is the command for listing nvm versionssome-command 2>&1 | yo why did this fail