Skip to content
Vitrino docs

Vitrino documentation

Vitrino runs Claude’s coding agent on your machine and replaces the terminal output with a transcript you can read. These pages cover installing it, using it, and how Claude Code’s own concepts appear inside it.

Getting started

  • InstallDownload and install Vitrino on macOS, Windows or Linux, and get past the first-run warnings.
  • First runAdd your API key, open a project folder, and send your first message.
  • UpdatingHow Vitrino tells you about a new version, and how to install it.

Using Vitrino

  • The transcriptHow a session reads: rows per event, collapsed detail, diffs, and what each part of a turn means.
  • PermissionsApproving tool use, what a permission card shows, and what “Always allow” saves.
  • Writing messagesThe composer: rich text, images, slash commands, and queueing work.
  • Projects and sessionsTabs, the session picker, resuming after a restart, and how sessions are stored.
  • Keyboard shortcutsEvery shortcut, on macOS and on Windows and Linux.
  • SettingsAccount, model and effort, appearance, and notifications.
  • TroubleshootingWhat to check when something does not work, and where the logs are.

Claude Code in Vitrino

  • ToolsThe tools Claude uses, and how each one is drawn in the transcript.
  • Subagents and background workNested transcripts, background tasks, and what happens while you wait.
  • Plans and questionsPlan mode, reviewing a plan, and answering Claude’s questions.
  • Skills and pluginsWhere skills, plugins and commands come from, and how they show up.
  • MCP serversMCP tools, prompts and elicitations, and how Vitrino presents them.
  • Slash commandsThe command palette, what each built-in command does, and which run locally.
  • Project configurationCLAUDE.md, settings files and permission rules — all read from your project, unchanged.