Save commands you understand
Snippets make repeated work easier, but a saved command is not automatically safe on every server. Use descriptive names and keep environment-specific addresses and assumptions visible. Do not put passwords, private keys or reusable access tokens into snippets.
- Open Settings → Snippets where available.
- Add a command you have reviewed and give it a meaningful name.
- Before using it, confirm the selected host, directory and account.
- Review the resulting command before execution, especially when it changes files or services.
A command that is harmless on a disposable server can be destructive on a production machine. Reusing a snippet should save typing, not eliminate the decision to run it.
History and completion
Autocomplete and ghost-text suggestions can help repeat earlier commands. Review them as input suggestions, not as evidence of what should run next. Change their behavior under Settings → Terminal → Interaction when the controls are available.
Shell-history import can bring existing commands into supported workflows. Choose the intended history file and check for errors after import. History can contain sensitive arguments and private paths, so do not treat it as ordinary public text or share it in a support request.
AI is a separate choice
History-based completion is not the same as asking an AI service to reason about a terminal. Configured AI or voice features can involve an external provider and selected context. Read AI assistance and voice before enabling them.
A suggestion never overrides server permissions or your responsibility to inspect the command. Keep backups and recovery options appropriate to the work you are performing.