Docs/Left sidebar/Objects
Objects
The Objects panel is your project seen by kind rather than by folder. Every object type you have — Book, Person, Meeting — appears with a count of how many notes are of that kind, and expanding one lists them. It's a way to find things by what they are, without having filed them together in the first place.
The list of types
Each row is one type: its icon, its name, and how many notes carry it. Types with no notes yet are still listed, so creating your first Book is something you can find rather than something you have to remember exists.
Click a type to expand it and see its notes; click again to collapse. Clicking a note opens it in the editor. Each note carries its own icon, which matters when a type has more specific kinds beneath it — a Book that's really a Novel shows the Novel icon while still being counted as a Book.
Opening a type as a collection
The ⤢ button on a type row opens all of its notes in the main area, where they're more than a list. There are three ways to look at the same set, plus a map for a type with a location, a board for a type with a choice field, and a timeline and a calendar for a type with a date:
Switching between layouts doesn't re-fetch anything; it's the same set of notes shown different ways. Clicking any row, card or pin opens that note — and Back brings you to the view again, just as you left it.
Changing notes from a view
You can retag notes and change their fields without leaving the view. Select a note by ⌘-clicking it (Ctrl-click on Windows and Linux), select a run of notes with ⇧-click, or press ⌘A for all of them; Esc clears the selection. A plain click still opens the note. Then right-click one of them (or press ⇧F10) for the menu, which acts on everything selected:
- Add Tag… asks for a tag, suggesting ones already in use, and adds it to each note.
- Remove Tag ▸ lists the tags the selected notes carry, with how many have each, and removes the one you pick.
- Set Property ▸ lists the type's fields; picking one opens just that field's editor — a date, a choice, a link to another note — and sets it on every selected note.
- Clear Property ▸ lists the fields the selected notes have filled in, with how many have each, and empties the one you pick.
- Edit Properties… opens every field at once.
On a timeline, right-click an event, or a note in its Undated list, for the same menu, for that one note. A timeline has no selection, so a click still just opens the note.
A field that links to another note — a project's owner, say — shows that note's name with a dotted underline, in a table or on a board's card. Right-click the name for a menu that acts on that note, not the row it's in: its first line names the note, then Add Tag…, Remove Tag… (offering the tags that note has) and Edit Properties… for its own fields. Your selection of rows stays as it was.
On a calendar, right-click any event (in the grid, a month or year band, the Undated list, or a day's list) for the same menu, for that one event. It also has Move to date…, which lets you pick a new day (see Moving an event to another day).
All of it works from the keyboard. Tab to a row or card (in a table, each row is a stop; ↵ opens the note) and press ⇧F10 or the menu key for its menu, then use the arrow keys and ↵. A note named in a field is a stop of its own: ↵, Space or ⇧F10 there opens that note's menu. When a change is made, a screen reader hears what changed (for example Tagged 3 of 3 notes with "active") instead of a dialog appearing; a dialog appears only if some notes couldn't be changed, listing which and why.
The view updates as soon as the notes do. These are ordinary edits to your notes, so local history keeps the earlier versions. A view inside a note has the same menu for one note at a time: right-click a row or card there, and it acts on just that note (a board in a note still doesn't move cards). An exported view is for looking at and has no menu.
The map
A location is written as latitude, then longitude, separated by a comma — 50.087, 14.421 is the middle of Prague. Each note with a location gets a pin; a note with the field empty, or with something that isn't a location, simply isn't on the map. The map frames itself to fit all the pins.
Each pin takes the colour of the note's own type, so on a Place map a restaurant and a hotel look different at a glance. Hover a pin to see a preview of the note, the same one a link to it shows: its title, its opening lines and the type's card fields. Click a pin to open the note, or right-click it to tag the note or set or clear its fields (see Changing notes from a view), in a map embedded in a note too.
The Auto, Light and Dark buttons at the map's top left choose its look. Auto follows Minerva's appearance; Light or Dark keeps that map light or dark whatever the appearance, so you can have a light street map in a dark window. The map changes at once and keeps its pins and where you were looking. The choice belongs to the view: it is kept with the tab, and Save as note and Copy as markdown carry it into the note. A map embedded in a note shows the saved look without the buttons.
The pins are your notes, but the streets and coastlines behind them are fetched from the internet as you look. Offline, the map has nothing to draw them on.
The Kanban board
A board lays a type's notes out in columns by one of its choice fields — a field with a fixed list of choices, an enum in Settings → Object Types: one column per choice, one card per note. Project's status makes columns for active, paused, done and abandoned, so a board of your projects needs no setup. If the type has more than one choice field, Group by in the view's header picks which one makes the columns.
The board is a built-in plugin, Kanban, which you can turn off in Settings → Plugins. Moves it makes are listed in Local History as Moved by Kanban.
The columns come in the order the type lists its choices, each headed with how many cards it holds. An empty column stays on the board, since it's still somewhere to put a card. A note whose field holds something that isn't one of the choices — status: blocked, typed by hand — gets a column of its own after the type's choices, rather than being hidden. Notes that leave the field empty gather in No value, at the end, which appears only when some note has no value.
A card shows the note's icon, its title and the view's visible fields, leaving out the one the columns already show. Within a column, cards follow the view's sort, or the order you've put them in (see Ordering cards). A filter on the grouping field limits the board to the columns you ticked; a filter on any other field leaves the columns alone and just thins the cards. Click a card to open its note. Rest the pointer on a card, or move to it with the keyboard, and a preview shows the note's title and the opening of the note, as when you hover a link to it, with the card's fields underneath. Esc closes it.
Moving a card
Drag a card to another column and the note's field changes to that column's choice, and the card lands where you drop it, between the two cards either side. Dropping it on No value takes the field out. Only that one line of the note changes, whether or not the note is open; if it's open in a tab with edits you haven't saved, you're asked whether to load the new version, as with any change made outside the editor. A drop on the spot the card already holds, or Esc while you're dragging, changes nothing, and finishing a drag never opens the note.
Without a pointer, right-click a card (or press ⇧F10 on it) and choose Move to ▸. It lists every column, including ones the board isn't showing, with the card's current column greyed out. Select several cards first — ⌘-click or ⇧-click — and it moves them all.
Each move is an ordinary save, so it shows up in Local History and can be restored from there like any other edit. While the board has focus, ⌘Z takes back the last move, and the one before it, and so on. If you've edited the note since, only the field is put back; if the field itself has changed again, the note is left alone and Minerva tells you so. ⌘Z on the board never touches the editor's own undo.
Ordering cards
Drag a card up or down its own column and drop it between two others: an accent line shows where it will land. Or press ⌥↑ / ⌥↓ on a focused card to move it one place. The order belongs to the view, like the column order. No note changes, so it doesn't appear in Local History, and ⌘Z doesn't take it back.
A card you haven't placed follows the ones you have, in the view's order, so a new note joins the bottom of its column. A card a filter hides keeps its place and comes back to it. A card that moves to another column keeps its place relative to the cards there that you've placed. A note that's renamed or deleted loses its place.
The order applies while the view isn't sorted. If you've sorted the view, by clicking a table's column header for instance, putting a card in order turns the sort off and keeps the board exactly as it was drawn, except for the card you moved. Sorting again later sets your order aside without losing it.
Arranging the columns
Drag a column's header sideways to put it somewhere else. Each header also has a ⋯ button (or right-click the header) with Move column left and Move column right. The order belongs to the view, not the type, so it never changes the type's list of choices, and choosing a different Group by starts the order afresh.
Show empty columns, beside Group by, is on to start with. Turn it off and columns with no cards drop out of sight; Move to ▸ still offers them.
By keyboard
The whole board is one stop on Tab, after the layout tabs. From there:
Moves, card order and column changes are announced to screen readers — Moved Garden Shed to done., Garden Shed moved above Roof.
Saving a board
Group by, the column order, the card order and Show empty columns belong to the view, like its sort and filters. They're kept with the tab, and Save as note (Minerva suggests a name like Project board) and Copy as markdown carry them into the note. A board in a note is for looking at: its cards open their notes, but they don't drag and have no Move to ▸, and the headers have no menus. A web page or PDF exported from the note, and a published site, show the same board read-only, its columns wrapping onto the page and each card a link to its note. Object views shows what a saved board looks like in the editor.
The timeline
A timeline puts a type's notes on a line of time. It's offered for any type with a date or datetime field, its own or one it inherits from its parent. Of the stock types, that's Event and Meeting (by date), Book and Article (by published), Project (by started), Idea (by created) and Claim (by As of). Minerva's own Meeting is a kind of Event, so a timeline of Events has your meetings on it, and a Meeting view can be a timeline of its own. A type you've made with a date field, or with a dated type as its parent, gets one too.
Event notes are read from their date and, if they have one, their end. Every other type is placed by its date field, one point per note — or, for a date like 1925, the stretch it names.
The timeline is a built-in plugin, Timeline, which you can turn off in Settings → Plugins. It only reads your notes: zooming and panning change what the view shows, never a note.
Date by
A type with more than one date field, such as a Trip with departs and returns, has a Date by picker beside the layout buttons, and the timeline places each note by the field you pick. A type with one date field has no picker. The first choice is date if the type has one, and otherwise its first date field. Event's end isn't offered on its own, because it ends the date range rather than starting one.
Only an Event's date and end draw bars. If you date an Event view by another field, or date any other type, every note is a point at that date, or the stretch an approximate date names, whatever its end says. Date by belongs to the view, like its range: it's kept with the tab and saved with the view. The calendar uses the same choice.
Points and bars
An event with only a date takes up the time its date names: a day, if you gave a day. An event with an end as well runs from its date to its end. How each one is drawn depends on how far you've zoomed in. Anything too short to see at the current zoom is a point, and anything wider is a bar, with its title inside the bar when it fits and beside it when it doesn't. A day-long event is a point when you're looking at a century and a bar when you're looking at a week. A time of day (2026-10-05T14:30) is a moment, so it stays a point.
Hover over an event, or move to it with the keyboard, and a preview shows the note's title and its opening lines, as when you hover a link to it, with the event's dates and its fields. Esc closes it. Click an event to open the note.
Dates that aren't exact
A date can be as coarse as you know it. 1969 means the whole of 1969 and 1969-07 the whole of July, so the event covers that stretch, drawn hatched with a dashed outline to show it's approximate. The hover preview says so, and a screen reader reads it out with the event. If an event has an exact start and an approximate end, only the end's stretch is hatched.
Years before 1 CE are written with a minus sign, counting the way astronomers do: 0000 is 1 BC and -0043 is 44 BC. That's one less than the BC number, because there's no year zero in BC counting. The timeline shows them as BC, in its axis labels too, so -0043-03-15 reads Mar 15, 44 BC. It can draw any year from 99,999 BCE to 99,999 CE.
Event's date and end are datetime fields, so they can carry a time of day as well. That's what puts two events on the same day in order: a 9:00 meeting is drawn before a 14:30 one, and the keyboard reaches them in that order. A date with no time covers the whole day.
Lanes
Events that would overlap, labels included, are stacked into lanes, one below the other, so nothing is drawn on top of anything else. Zooming in spreads events out, so they need fewer lanes. The lanes are worked out again on every zoom, but panning never moves an event to a different lane.
The Undated tray
Events the timeline can't place are listed in the Undated tray under it, each with the reason: No date, a date that couldn't be read (1969-02-30), or one outside the years it can draw. Nothing is left out without a word, and clicking one opens it so you can fix it. Click the tray's Undated heading to hide the list and give the drawing the room; the heading stays, with its count, and clicking it again brings the list back.
An end that comes before its date, or that isn't a date at all, doesn't send the event to the tray. The event is drawn at its start as if it had no end, and its hover preview flags the end that was ignored.
Zooming and panning
The timeline opens fitted to all its dated events. Scroll or pinch to zoom in or out around the pointer. Drag, scroll sideways, or hold ⇧ while scrolling to pan. The − and + buttons above the drawing zoom around the middle, and Fit shows every dated event again. If the view has a range filter on the field it's dated by, Fit stays within it. Finishing a drag never opens the event under it. The axis labels follow the zoom, from years down to hours and minutes.
The range you're looking at belongs to the view. It's kept with the tab, so switching to the table and back, or going Back to it, brings you to the same stretch of time.
By keyboard
The drawing is one stop on Tab. From there:
Each event is announced by its title and dates, such as Moon landing, Jul 20, 1969, and a bar's description includes its end.
Saving a timeline
The visible range goes with the view: Save as note (Minerva suggests Event timeline) and Copy as markdown keep it as a from and to, so the note shows the same stretch of time, and a Date by choice as dateBy. A timeline still fitted to its events saves no range and fits itself whenever it's shown. A timeline in a note can be stepped through and opened, but not zoomed or panned. A web page or PDF exported from the note, and a published site, show a still timeline of the saved range, dated by the same field, with each event a link to its note, approximate dates marked as approximate, and the undated events listed after it. A very crowded timeline is drawn with its first 30 lanes and a note of how many more events there are. Object views shows what a saved timeline looks like in the editor.
A timeline is for seeing how things fall across time, whether that's hours, years or centuries, and how they overlap. A calendar is for seeing what happens on which day, a month at a time. Both read the same dates in the same way, so you can switch between them freely. The calendar shows months only: there's no week or day view with the hours down the side yet, and no repeating events. Each note is one event, on the dates it gives.
The calendar
A calendar puts a type's notes on the days of a month. It's offered for the same types as the timeline: any type with a date or datetime field, its own or inherited. A calendar of Events has your Meetings on it, a calendar of Books shows them by when they were published, and a type you've made with a date field gets one too. If the type has more than one date field, the same Date by picker chooses which one places the notes, and the choice carries over when you switch between the timeline and the calendar.
As on the timeline, only an Event's date and end make an event that lasts several days. Every other type, and an Event dated by another field, puts each note on the one day its date names.
The calendar is a built-in plugin, Calendar, which you can turn off in Settings → Plugins. When you move an event to another day, Local History lists the change as Moved by Calendar.
The month
The calendar opens on the current month. The ‹ and › buttons beside the month's name go back or forward a month, and Today comes back to this one. Today's date is ringed, with its number in bold. A month takes four, five or six rows, as many as its weeks need, and the days of the months either side that fill out the first and last rows are dimmed. Their events are still shown.
The month you're on belongs to the view. It's kept with the tab, so switching to the table and back, or going Back to it, brings you to the same month.
Events on their days
An event with only a day sits in that day's cell. One with a time of day (2026-10-06T09:30) sits there too, with the time in front of its title, and a day's events come in order: anything that began on an earlier day first, then by time, then by title.
An event with an end on a later day is a bar across the days it covers. A bar that runs past the end of a week carries on at the start of the next row, with ‹ and › marking where it's been cut. Each bar keeps one line across its row, so the events under it don't jump about from day to day. If the end is known only to the month or year (end: 2026-11), the days it might cover are hatched with a dashed edge, as on the timeline, to show that they're approximate.
Rest the pointer on an event, or move to it with the keyboard in a day's list, and a preview shows the note's title and opening lines, as when you hover a link to it, with the event's dates and the type's card fields. Esc closes it. Click an event to open its note.
Times and time zones
A time written on its own, like 09:30, is a time on the clock: the event is at 9:30 wherever you are, the way a dentist's appointment is. A time written with an offset from UTC, like 2026-10-06T23:30-07:00 or a Z at the end, is a fixed moment, so the calendar shows it at that moment in your own time zone. An evening call written in California can fall on the next morning in Europe, and the calendar puts it on that next day, which is the day it happens for you. When that moves it to a different day from the one written, its hover preview says Written as 2026-10-06T23:30-07:00, so the value in the note is never hidden. Datetime fields describe both forms.
Dates without a day
A date can be as coarse as you know it, and the calendar doesn't pretend otherwise. 2026-10 means some time in October, so there's no one day to put it in. Instead it goes in a band above the grid, headed October 2026 · no day, and a date like 2026 goes in a band headed 2026 · no month. A month-only event is listed on every month page its range covers, and a year-only one on every page of its year. Where the range runs past the month you're on, the entry is marked since September or until December. That way a festival dated only to the summer shows up on June, July and August's pages, and never pretends to be on the 1st.
The bands sit above the grid but come after it when you Tab through the view, since the grid is the main thing you came for.
The Undated tray
Notes the calendar can't place are listed in the Undated tray under the grid, each with the reason: No date, or a date that couldn't be read (2026-02-30). Clicking one opens it so you can fix it. As on the timeline, click the Undated heading to hide the list and give the month the room; the heading keeps its count, and clicking it again brings the list back.
A full day
A cell holds as many events as fit at the height the window gives it. When there are more, the cell ends with +3 more, and clicking it opens the day's whole list, in the same order as the cell. Pressing ⏎ on a day opens the same list, so every event can be reached by keyboard however full its day is. In the list, ↑ ↓ Home End move between events, each showing its preview, ⏎ opens one, and Esc closes the list and puts you back on the day.
Moving an event to another day
Drag an event onto another day and it moves there. While you drag, the days it will cover are highlighted. A bar moves as a whole, from wherever you picked it up: drag the third day of a three-day trip onto the 20th and the trip ends on the 20th. Its end moves by the same number of days as its date, so it lasts just as long, and a time of day stays as it was. Esc while you're dragging changes nothing, and finishing a drag never opens the note. You can drop on any day you can see, including the dimmed ones from the months either side.
Or right-click the event and choose Move to date…. Without a pointer, open the day's list, move to the event, and press ⇧F10 (or the menu key) for its menu. The menu also tags the event and sets or clears its fields (see Changing notes from a view). Focus goes back to the event's own day in the grid, and a bar above the grid says what's happening. Move to another day with the arrow keys or Page Up and Page Down, as usual, and press ⏎ to move the event there, or Esc to leave it where it is. Clicking a day works too. For a day months or years away, type it in the bar's box instead, as 2026-12-24, and the calendar turns to that month.
A move writes the new date and end into the note in one save, in the same form they were written in, and changes nothing else. An open tab of the note picks up the change; if it has edits you haven't saved, you're asked which version to keep, as with any change made outside the editor. A time written with an offset keeps that offset, and if a change to or from summer time falls between the old day and the new, the time is adjusted so that the event still lands on the day you dropped it on.
Each move is an ordinary save, so it shows up in Local History and can be restored from there like any other edit. While the calendar has focus, ⌘Z takes back the last move, then the one before it, and so on. It only undoes moves made on a calendar, never a board's. If you've edited the note since, only the dates are put back; if the dates themselves have changed again, the note is left alone and Minerva tells you so.
What doesn't move. An event whose date or end is only a month or a year has no day to move by, so it can't be dragged or moved to a date. Minerva won't guess a day for you and write it into your note. Trying it tells you why, as in Launch ends 2026-11, which has no day to move; open it to edit., and in the menu Move to date… is greyed out with the same reason under it. Open the note and change the date there. Events in the bands are the same.
By keyboard
The grid is one stop on Tab, on the day you were last on: to begin with, today, or the 1st on another month's page. From there:
Each day is announced with its date and how many events it has, such as Tuesday, 6 October 2026, 2 events, and moves are announced as they happen.
Which day a week starts on
The first column is the day your weeks start on. To begin with, Minerva takes it from your computer's region: Monday in most of Europe, Sunday in the United States. To choose for yourself, open Settings → Appearance and set Week starts on to Monday, Sunday or Saturday; Automatic goes back to your region's, and names the day it means. In the same place, Show week numbers adds a column of ISO week numbers down the left, the numbering most of Europe uses, where week 1 is the week with the year's first Thursday in it. Both settings belong to this computer rather than to the thoughtbase, so a colleague who opens the same thoughtbase sees weeks the way they've set them.
Saving a calendar
The month you're on goes with the view: Save as note (Minerva suggests Event calendar) and Copy as markdown keep it as a month, so the note opens on that month, and a Date by choice as dateBy. A calendar you've brought back with Today saves no month, and always opens on the current one. In a note, a calendar can still be paged from month to month, and its events opened, but nothing in it can be moved; open the type's calendar for that. A web page or PDF exported from the note, and a published site, show the saved month (or, with none saved, the month it was exported in) as a still grid, with every event listed in its day, however many there are, and long titles written out in full. A busy day just makes a taller row, and a week is never split across two pages. Each event is a link to its note, approximate days are hatched with a key underneath, and the bands and the Undated list follow the grid. The week starts on the day your own settings say, as in the preview you exported from. Object views shows what a saved calendar looks like in the editor.
Just one folder's objects
To see only the objects in part of your thoughtbase, right-click a folder in the Notes tree and choose View Objects. It lists each type found in that folder and everything under it, with how many there are — plus the more general types they belong to, so a folder of restaurants offers both Restaurant and Place. Pick one and its view opens showing only that folder's notes; the tab is named for the folder (Place · trip/prague). The view opens as a table; switch layouts as you like. The in trip/prague ✕ chip in the view's header widens it back to the whole thoughtbase. The thoughtbase's own menu — right-click the top row of the Notes tree, or the empty space below it — has View Objects too, covering your whole thoughtbase.
If you rename or move the folder later, its views follow it: views saved in notes and view tabs you have open now point at the folder's new place, so they keep showing the same notes.
Filtering by a field
Use Filter in a view's header to narrow it by one of the type's fields. For a text, choice or link field, tick the values you want from the ones that actually appear — each shown with how many notes have it — so a map of places can show only those whose city is Prague. Location fields aren't offered. For a number or a date, give a lowest value, a highest value, or both; a date range can be as coarse as a month (2026-05) or a year, and each end takes in the whole of what it names, so a highest value of 2026-05 keeps every day of May. Dates are compared as points in time, so times written with different time zones, and years before 1 CE (-0043 is 44 BC), fall in the right order. Sorting a date column works the same way. Filters on different fields all apply at once, and a note with the field left empty is left out. Each active filter shows as a chip: click it to change it, or its ✕ to remove it. The count then reads 2 of 3, so you can see how much is hidden.
If a type has a parent — Novel's parent being Book — then asking for Books gives you the Novels too. The count on the Book row includes them, and so does its table.
To narrow such a view to some of the specific types, use Filter → Type. It's there only when the type has more specific ones, and lists them as a tree — Restaurant with Pizzeria under it, Museum — each with how many notes it would keep. Ticking Restaurant keeps the restaurants and the pizzerias, since a pizzeria is a kind of restaurant, so a map of places can show just where to eat. Like any filter, it's saved with the view.
To narrow a view by tags, use Filter → Tags. It lists the tags the view's notes carry, each with how many have it, and counts tags from a note's frontmatter and its #tags in the text alike. Tick one or more, then choose how they combine: Any keeps a note with at least one of them, All one with every one, and None one with none of them. A tag also takes in the tags nested under it, so #ml covers #ml/nlp, and None of #archived hides #archived/2025 as well. Tags are matched without regard to case. Like any filter, it's saved with the view and comes along when you save it as a note or copy it as markdown.
Saved views
Once you've arranged a type's view the way you like it — a map of places, or a table sorted by rating with three columns showing — use Save as note to keep it as a note. You're asked for a name (Minerva suggests one, like Restaurant map), and a new note is created at the top of your thoughtbase with the view embedded in it, live: it shows the current objects every time you open it, arranged exactly as you saved it. Because it's an ordinary note, you can move it, rename it, link to it, or add your own writing around the view.
To put the view into a note you already have, use Copy as markdown instead and paste it where you want it. What's copied is the same live view that Save as note embeds — the same layout, sorting, columns, folder and filters — so it shows the same thing.
On a table, Export CSV saves the notes it shows as a spreadsheet file: a Title column and one column per visible field, in the table's order. Dates are written in their ISO form (2026-10-05T14:30) rather than as shown, so another program can read them back.
A view saved from a folder keeps that folder, and a filtered view keeps its filters. In the note's preview, in a web page or PDF exported from it, and on a published site, you see just those objects — the filter controls and the folder chip belong to the view's own tab, so they don't appear there. A Markdown export keeps the view as the text you'd see in the editor. To change what a saved view shows, open the type's view, arrange it again, and save it anew. Object views has more on views inside notes.
Views saved before Save as note used to be listed under their types in this panel. They no longer appear. To get one back, set the view up again and use Save as note.
Excerpts
Below your types sits Excerpts — the passages you've saved from sources. It isn't an object type you defined, but it belongs here for the same reason: it's a kind of thing you accumulate, and you often want to browse all of it at once.
Where types come from
The types listed here are defined in Settings → Object Types, where you can add fields, change an icon, or create your own. A note becomes one of them by carrying a type: property — see Typed notes.