Assistant

Explain a command or an error

An optional assistant explains a command or an error with Claude (Anthropic). It is off by default. To use it, turn it on in Settings › Assistant and save your Anthropic API key (kept in the vault). Then select text in a terminal and right-click › Explain the selected command or Explain the selected error: a window shows exactly the text that will be sent, which you can edit, and nothing leaves until you press Send. Only that text is sent, with your key; remove passwords, tokens or internal names first. Answers come from an AI model: check them before running what they suggest.

All of it is yours to configure in the same page: the model (any model ID your key can use; Load models lists them and also checks the key), the effort (from low, fastest, to max, slowest, or the model's default), whether the model reasons before answering, whether Anthropic may answer with another model when the chosen one declines (turn it off for gateways or models that do not support it), the longest answer in tokens, the endpoint (empty for Anthropic's API, or a compatible company gateway: https, or http only on localhost; your key is sent there), and your instructions added to every request (for example your distribution and shell). Restore defaults goes back to Claude Opus 5.5, medium effort, fallback on. The settings are kept in the vault with the key. The window with the answer shows which model answered.