First Run

The first launch walks through five visible stages:

  1. Welcome introduces Kucedr as a personal agent for everyday tasks, coding, and background work.
  2. Account lets you sign in, create an account, or continue in local-only mode.
  3. Model requires at least one saved model-provider API key. A key that has only been typed but not saved does not complete this stage.
  4. Search optionally connects a supported search provider.
  5. Models selects the main Assistant provider and model. Realtime voice, transcription, speech, image, audio, video, and search choices are optional.

Once an Assistant provider and model are stored, Kucedr skips setup on later launches and opens straight into chat. Everything chosen here can be changed later in Settings, and the provider and model for scheduled work and health checks are configured separately on their own Settings pages.

Setup checks that a provider and model have been saved; it does not test whether your credentials can complete a request. Send a short message in Home. If it fails, check the saved key, selected model, network access, and provider error.

After Setup

  1. Start chatting. Review any permission card for file or command actions outside trusted locations before allowing it.
  2. Attach files through the composer, or dictate instead of typing. The composer does not check file count, type, size, or model support before sending, so an unsupported attachment may fail during processing.
  3. Configure speech-to-text, text-to-speech, and image, video, and audio generation if you want them.
  4. Extend Kucedr with Plugins, Apps, Channels, or Remote Agents, according to the work you want to do.
  5. Build a focused in-app tool with the Kucedr SDK when a workflow benefits from its own interface.

Finding Your Way Around

Settings groups its sidebar into General (including Providers, Library, and Workspace), Assistant, Brain, and Extensions (Plugins, Apps, Channels, and Remote Agents). Agent links and route search expose deeper pages such as Skills, Tasks, Chat History, Knowledge, Health, and Permissions. Coding is marked Soon; its old route opens Workspace.

  • Cmd/Ctrl+F opens a route and setting search palette.
  • Cmd/Ctrl+/ focuses the prompt editor.
  • Deep pages use breadcrumbs, unknown routes show a recovery view, and page transitions respect your reduced-motion preference.

The Search settings page configures the same Brave or Tavily engine and key the agent's web-search tool uses. It is not a separate local-search feature.

Application Preferences

View the app name and version, toggle the tray or menu-bar icon, keep the computer awake while still letting the display sleep, open the application-data folder, pick English or Italian, and choose a light, dark, or system theme. System mode follows OS theme changes.

Media Permissions

System pages cover Microphone, Camera, and Screen capture. On macOS, Kucedr shows permission status, can request access, and opens the relevant System Settings pane; microphone recording, camera preview, and screen capture can each be tested and played back.

On other platforms the explicit system permission status can report as unknown, and application-level microphone and camera toggles do not currently disable capture. Display capture uses the macOS system picker where available and a native Kucedr source chooser when Electron returns multiple sources; some Linux PipeWire environments expose only one portal-managed source.

Platforms And Languages

PlatformPackaging
WindowsNSIS installer for x64, selectable install directory, desktop shortcut, app data retained on uninstall.
macOSPKG and DMG for x64 and arm64, dark-mode support, hardened runtime, microphone and camera entitlements.
LinuxAppImage and DEB.

The interface ships English and Italian catalogs with a locale selector, and light, dark, and system themes. Localization is partial: major first-run and home-screen copy, including the prompt suggestions and editor placeholder, is still hardcoded in English.