Search & Command Palette

Press ⌘+F for global search and command execution from anywhere.

Opening

Where	Shortcut
Anywhere in the app	⌘+F
Global hotkey (if set)	Configured in Settings → Keyboard Shortcuts
Sidebar Search button	Click

Esc closes the palette.

Scope

Type to search across:

• Notes — title and body
• Journal entries — title and body
• Tasks — title, description, project name
• Inbox items — title, source URL, captured text
• Tags — direct match by tag name
• Settings — jumps to a settings panel by name

Results are grouped by type, with snippet previews and relative dates.

Scoped Search

Number prefixes scope to a single source:

Prefix	Scope
1	Notes only
2	Journal only
3	Tasks only
4	Inbox only

Type the number, then your query.

Tag Filter

Type #tagname anywhere in your query to filter results to items tagged with that tag. Multiple tags compose with AND.

Examples:

• #research neural networks — items tagged research containing "neural networks"
• #daily 1 review — notes only, tagged daily, containing "review"

Recents

When the palette is empty, recent queries and frequently used searches appear. Pin a query (right-click) to keep it at the top.

Semantic Search

If embeddings are enabled in Settings → AI, search results are ranked by both keyword match and semantic similarity. This means:

• Queries phrased differently from the source can still find it
• "Setting up authentication" matches notes about "OAuth flow" even without keyword overlap
• Older notes resurface when their meaning matches your current query

Toggle semantic search per query with the search bar's options menu, or always-on in settings.

See Embeddings & Semantic Search for setup.

Result Actions

For each result row:

• Enter — open in current tab
• ⌘+Enter — open in new tab
• ⌘+⌥+Enter — open in split pane
• → — preview without opening

Performance

Search runs against the index DB for keyword match. Embedding-based ranking happens on the device, not the server. Even with semantic search enabled, queries don't leave your machine.

For large vaults, the keyword index is FTS5-backed and stays sub-100ms even with 10k+ documents.