Open Editor settings from the header, command palette, or Ctrl/Cmd+, with the VS Code keymap. The window fills the screen on phones. Search settings or choose Appearance, Writing, Preview, Keyboard Shortcuts, or Advanced. Preferences are stored in this browser; these controls do not change your published site’s configuration.

Appearance includes source text size and the existing site palette controls when available. Writing includes line wrapping and the editor keymap. Preview controls layout and synchronized scrolling. Keyboard Shortcuts contains the application preset, search, conflict-aware shortcut recorder, and reset controls.

Find files and commands #

Ctrl/Cmd+P finds files; Ctrl/Cmd+Shift+P finds commands with the default application keymap. Both support fuzzy matching. The file picker searches filename, path and post title. Recent choices appear first on an empty query. With a query, history breaks ties between similarly strong matches; exact matches always take priority. Command history is local to this browser. File history is scoped to the workspace. Neither stores your search terms or draft contents.

Advanced settings lets you disable remembering picks or clear command history and recent files for the current workspace. Blocked browser storage does not prevent opening files or executing commands; history then lasts only for the current tab.

Diagnostics without sidebar clutter #

Use Diagnostics in the footer or Toggle Performance Diagnostics in the palette. Startup indexing details remain available in the drawer. Copy full report includes startup status and up to 100 recent samples per measurement. Agents can read window.MarkataDiagnostics.snapshot() directly. Missing samples are not evidence of instant completion.

file_open ends when the editor receives the file; save_acknowledgement ends when the server acknowledges a write, not when a deployment completes. edit_to_preview measures the accepted preview update, including debounce and render time. file_request and publish_request measure resource requests only: they are not proof of a durable save, successful publication or deployment. Search timing excludes result DOM rendering. Validate end-to-end latency on a real site before making performance claims.

See the Go workbench guide for publishing, Changes review, and comparing unsynced copies.