PlyLoom / User guide

PlyLoom user guide

Everything PlyLoom can do, from the general to the particular: how a map is built, working on a sheet and pictures, the passe-partout of neighbouring sheets, views, the story and reading, dictionaries and the filter, files, and ready-made recipes. First steps are on the Getting started page.

The same guide is inside the app: Help → User guide.

Basics

What a map is made of and where things are on the screen.

How a map is built

A PlyLoom map is a project made of sheets. A sheet answers one question, its nodes are units of meaning and relations join them. The same entity can stand on several sheets at once.

Project
The whole map: sheets, nodes, relations, dictionaries, saved views, assemblies and the story. It is kept in one .plyloom file and, while you work, in the browser too.
Sheet
One side of the topic, one question: Requirements, Architecture, Risks. A sheet has a name, colour, type, tags and a description. There is no node limit.
Node (entity)
A requirement, component, idea, person or risk — any unit of meaning. A node has one name, a type, a text, attributes and, if you like, a picture.
Appearance
Where a node stands on a particular sheet. A node can stand on several sheets: each has its own position, while the name, text and attributes are shared.
Relation
An arrow from one node to another with a type (“depends on”, “supports”…) and an optional label. It may join nodes on different sheets.
Dictionaries
The map's shared lists: sheet, node and relation types, sheet tags and attributes — node fields such as Status or Due date.

Three kinds of lines

  • Solid grey — a relation inside a sheet.
  • Dashed purple — a relation to a node on another sheet: to its card on the passe-partout or into the other half of a spread.
  • Gold — the same entity on different sheets. Nobody draws it: it follows from the sheets the node stands on.

The rule of a sheet: when a sheet answers two questions, divide it — Sheet properties → Lay the sheet out by a key does it for you. Size alone is no reason: a large sheet reads well with the filter and search.

See also: Sheets: create, set up, divide, One entity on several sheets, The map's dictionaries

The screen and panels

At the top, two rows — menus and views — and the sheet bar under them. The Nodes panel on the left, Properties on the right, the sheet in the middle. The canvas buttons sit in its lower right corner.

The window: menus and views at the top, the sheet bar, nodes on the left, the sheet in the middle, properties on the right.
The window: menus and views at the top, the sheet bar, nodes on the left, the sheet in the middle, properties on the right.

Row one: menus

Logo
Takes you back to work — from the guide, the contents or a window.
Project
New project, open and download .plyloom, save to a file, version history, assemble a document, a table or a diagram, compare with a file, import, export and examples.
AI
The AI window: a new map from materials or refining this one; paste an answer, get back to the edits.
Dictionaries · Settings · Filter
The map's shared lists; how cards, the passe-partout and the counts look; search and filtering, saved views.
Story
Watch the story of the map, add a step, edit the steps.
Help
Getting started, the guide, the guide as a map, working with AI, keyboard shortcuts, what's new, the project check.
Reading
The same map without editing — to show it to someone.
↶ ↷, the name, RU | EN, ⌃
Undo and redo; the map's name (click to rename); the interface language; ⌃ folds the top so only the sheet bar stays (Ctrl+.).

Row two: views

Sheet, All sheets, Spreads, Stack, Contents, Architecture, All-to-1. On the right of this row are the settings of the view on screen: which links to show, the line legend, the look of the stack and so on.

The sheet bar

Always on screen: “← 3/8 →” pages through the sheets (Alt+← and Alt+→), then the sheet tabs (double-click to rename), + for a new sheet, the counts “sheets · nodes · attributes” (click for Settings) and the saving state (click for the version history). With the top folded, the view is chosen here too, and ⌄ brings the top back.

Panels and canvas buttons

Nodes
The palette of types: drag a type onto the sheet or click it; + Node… opens a form with the name, type and sheet.
Properties
The sheet on top, the selected node below, the project check last. Sections fold from their headers.
Canvas buttons
In the canvas's lower right corner: ↩ Back (it appears after a jump to a node), − 100 % + for zoom, Fit, Optimize with its ▾ menu, Unfold all ▾, Snapshot ▾ for a picture of what the canvas shows, and “?” for the keyboard shortcuts. The set of buttons is the same in every view; a button that is not available in a view is greyed out, and its tooltip says why. With the passe-partout unfolded the buttons stand inside the sheet as signs only (the words are in the tooltips), so they never cover a card.
Passe-partout ▾
In the upper left corner of the passe-partout, on a tab of its own: the neighbouring sheets around the sheet — unfold, or fold to strips. No card ever lies under the button.

The panels fold into narrow strips with the ‹ and › arrows to give the sheet the whole width. On screens narrower than 1024 pixels they open from the ☰ and ⓘ buttons in the sheet bar.

See also: Phones and tablets, Keys and gestures, The passe-partout: unfolded and folded

Back to sections

Working on a sheet

Nodes, pictures, relations and layout — what you do every day.

Nodes: create, read, edit

A folded card shows its name; ↘ in its lower right corner unfolds it in place. The name is edited as in draw.io — select and type; the picture, the text and the attributes are edited right in the unfolded card.

Create

  • Drag a type from the Nodes panel onto the sheet: the node appears where you drop it. Clicking a type adds a node to the current sheet in a free spot.
  • Double-click empty space on the sheet: a New node appears.
  • + Node… in the panel opens a form: name, type, sheet and a note, which becomes the node's text.
  • Drag from a dot on a card's edge and release on empty space: a node of the same type appears, already connected.
  • Drop a picture file on empty space: a node with it appears, named after the file.

The panel has the groups Basic, Project types, Flowchart, BPMN and Canvas; Find a type… searches by name. A new node is named after its type — just start typing to give it a name.

The name, as in draw.io

  • Select a card and start typing: the name is replaced by what you type. Enter keeps it, Esc brings the old one back.
  • Enter or F2 on a selected card edits the name without erasing it.
  • A double-click on a card selects the whole name — to copy or replace it.

Unfold and read

↘ in the lower right corner unfolds the card in place, from its upper left corner down to the lower right: it grows wider and shows the Name, then the sections Picture, Text and Attributes; each section folds from its own header. ↖ folds the card back. The canvas moves by itself so the unfolded card is seen whole, and a new field — an attribute just added, say — never hides under the passe-partout or the buttons. Neighbours move aside on screen so nothing is covered; positions in the file do not change. Unfold all ▾ among the canvas buttons unfolds every card on screen — pictures only, text only, attributes only, or everything — and folds them back. Unfolded cards stay unfolded when you switch views: in spreads, in All-to-1 and in the Contents list. In an unfolded card the name is larger than the text. Empty sections (Picture, Attributes, an empty text) show only on the selected card, where they can be filled.

Below 60 % zoom the text of unfolded cards is smaller than 8 pixels on screen and cannot be read anyway. The name is then drawn large, 11–13 pixels on screen, over the top of the card, and the text and the picture stay under it as they are. A long name wraps by words and, if it still does not fit, ends with an ellipsis. The card keeps its size. Find the card by its name and zoom in to read the text.

Unfolded cards: sections fold one by one, neighbours moved aside.
Unfolded cards: sections fold one by one, neighbours moved aside.

Edit

A click on the text of an unfolded card edits it in place: the field grows with the text, and Esc or a click outside ends it. Selecting text with the mouse does not start editing, so it can be copied. A click on the attributes opens the same fields as in Properties, with + attribute. Everything is saved at once and shared by every sheet the node stands on.

The text is edited right in the unfolded card.
The text is edited right in the unfolded card.

Properties on the right has everything about the node at once: name, type, privacy, the sheets it stands on, text, picture, attributes and related nodes. The node type changes there too; the card's shape follows the type and its size stays.

In the textResult
**bold**bold
lines starting with - a list (one level)
| A | B |, then |---|---| and rowsa table; the first row is the header
an empty linea new paragraph

There are no headings, links, italics or numbered lists. A name holds up to 200 characters, a text up to a million. A node's picture is separate from its text — see Pictures in nodes.

Delete

Select a node and press Del (⌫ on a Mac) or Delete node in the properties: the node goes together with its relations. If it stands on several sheets, PlyLoom asks first — deleting removes it from all of them. To take a node off one sheet only, press × next to that sheet in the properties. Undo brings everything back.

See also: Pictures in nodes, Relations, Node attributes, One entity on several sheets

Pictures in nodes

A picture sometimes explains better than a paragraph. A node can have one picture: small on a folded card, larger in an unfolded one, full screen on a click.

Adding one

  • Drop a file on a card: the picture goes into that node. On empty space: a new node appears, named after the file; several files stand side by side in a row.
  • Ctrl+V with a picture on the clipboard (a screenshot, an image copied in a browser): with a node selected it goes into it; with nothing selected, a new node appears where the pointer is.
  • An unfolded card, the Picture section: Add a picture, and on a card with a picture Replace… and Remove. The same buttons and the caption are in Properties → Picture.

What happens to it

The picture is made small at once and the original is not kept: 640 points on the long side for an unfolded card and documents, and 192 for a folded card, as WebP (JPEG or PNG where the browser does not write WebP). A phone photo is turned the right way up, and the photo's metadata — place, phone model — are gone after compressing. PNG, JPEG, WebP and GIF (the first frame) are accepted, up to 25 MB and 40 megapixels. SVG is not accepted: it can carry code.

Caption
What the picture shows, in your words. Screen readers read it, Markdown documents and AI get it instead of the picture.
Settings
Settings → Cards → Pictures on folded cards hides and shows the small pictures at once. Unfold all ▾ → Pictures unfolds only the cards that have one.
In the file
One picture on five nodes is kept once. A map with pictures is written as version 4, which PlyLoom 0.14 and later open; a map without pictures stays version 3.
Size
Settings → Counts shows the number of pictures and the size of the map, text and pictures apart. The size also shows in the version history and in the menu at Download .plyloom. Past 25 MB of pictures PlyLoom calmly warns once.

Where pictures go and where not

  • Word, HTML and PDF get the picture with its caption; Markdown gets the caption; CSV and Mermaid nothing; draw.io a small picture inside the shape.
  • A picture never goes to an AI — only “[picture: caption]”, and only for nodes that are not private.
  • A private node's picture is private: a copy without the private part does not take it.
  • Ctrl+C and Ctrl+V carry a card to another map with its picture. Compare with a file shows picture added, replaced, removed.

See also: Nodes: create, read, edit, Assemble: a document, a table, a diagram, Private: what never leaves the map

Relations

A relation is drawn from a dot on a card's edge to another node — on the same sheet, in the other half of a spread or on the passe-partout.

  1. Hover a card: four dots appear on its edges.
  2. Drag any dot to another node. A highlighted target means the relation will take. The dot is only a grip: the line picks its side by itself and re-routes when nodes move.
  3. Release over a node or a card on the passe-partout: the relation is created with the type “reference”. Release on empty space: a new connected node appears.
  4. Click the line: the relation is selected and a panel opens next to it with its type, label, arrow and Delete relation.
A relation is dragged from a dot on a card's edge to another card.
A relation is dragged from a dot on a card's edge to another card.

Another way is + connect to… in the properties of the selected node: then click a node or a card on the passe-partout; Esc cancels.

Built-in types: then, depends on, supports, contradicts, reference, inherits from, implements, uses, answers. Add your own in Dictionaries → Relations.

Change and delete

Del or the button in its panel deletes the selected relation; Esc clears the selection. “With an arrow” makes the relation directed or not. The same is in the node's properties: Related nodes → ✎ on a row, or ×. Relations to nodes on other sheets are edited there too.

Every “contradicts” relation appears in the project check, so disputed points are listed.

See also: Nodes: create, read, edit, Project check

Moving, zoom and layout

The sheet pans and zooms inside the passe-partout. Fit shows all of it; Optimize rearranges the cards to suit the window.

Pan
Drag empty space on the sheet or turn the mouse wheel. The wheel scrolls vertically and Shift + wheel horizontally; a trackpad scrolls in any direction. It works the same in All sheets, the stack, the Contents diagram and Architecture.
Zoom
Ctrl + wheel (⌘ on a Mac), a two-finger pinch, or − and + among the canvas buttons; from 15 to 250 %.
Move a node
Drag the card. Dropped in the other half of a spread, it stands on that sheet too.
Fit
Picks a zoom so the whole sheet fits, and lays the passe-partout out afresh.
Snapshot ▾
A picture of what the canvas shows now, twice the size of the screen: Copy the picture puts it on the clipboard for a chat, an email or a slide; Download PNG saves a file. The canvas buttons stay out of it. In the corner is a small half-transparent mark, the logo and “PlyLoom”. For the whole sheet, press Fit first. The whole stack or architecture, SVG included, comes from As a picture ↓.

Optimize

Rearranges the cards so the sheet fits the window at a zoom of at least 90 %, lines do not pass through cards and cross as little as possible, and lines are short. Cards never overlap. Groups of related cards are laid out separately and then placed together to match the shape of the window; cards without relations fill the free space. Frames and lanes move together with their contents. It works on a sheet, in each half of a spread and in All-to-1. With more than 120 cards on screen the calculation runs in the background and can be cancelled. When it finishes, you see how the zoom, the lines through cards and the crossings changed. Undo restores the previous positions in one step.

The ▾ menu next to the button lets you choose a layout method. The chosen method is applied even if the previous layout scored better.

Automatic
The same as the button itself: tries each method and keeps the best. If none is better than the current layout, the cards stay where they are.
Left to right · Top to bottom
Cards stand in layers following the relations: where a relation starts is left of (or above) where it ends. A long chain wraps onto the next row.
Around the central node
The node with the most relations in the middle, its neighbours on its sides or in a circle.
Grid
Rows and columns matching the window; related cards stand next to each other.
Keep the order
The cards keep their order: overlaps and large gaps are removed and the sheet is fitted to the shape of the window. Good for a sheet laid out by hand.

Optimizing drops the bend points of lines on those sheets and gives groups moved by hand on the passe-partout their automatic places back.

Large sheets

At a small zoom, sheets over 200 nodes show cards without text, and over 500 a simplified picture (Simplified view). Select a node, find it with search or press Zoom in, and the regular cards come back.

See also: The passe-partout: unfolded and folded, Spreads, Jumping to a node

Jumping to a node

A jump opens the node's sheet and shows the node so it is easy to read. You can jump from a card on the passe-partout, “+N”, a relation in Properties, the search, the project check, or with a double-click in the overviews.

Zoom
The node ends up in the middle of the window at a zoom of at least 90 %. If the whole sheet fits at that zoom, you see the whole sheet; if not, the node with its related nodes; if those do not fit either, the node alone. The view moves to the node smoothly; the mouse wheel or a click on empty space stops it.
Unfolded cards
If the card you jumped from was unfolded, or Reading is on, the node's card unfolds too, with the same sections open. On the next jump it folds again unless you changed it (opened a section, edited the text or attributes). A jump from a folded card unfolds nothing.
↩ Back
Appears to the left of the canvas buttons after a jump. It restores the view, the sheet, the selection, the zoom and the unfolded cards to what they were before the jump. It remembers up to 40 jumps; its tooltip says where it leads.

In a spread, the node's sheet opens in the same half. In the stack, the node's plate becomes active.

See also: The passe-partout: unfolded and folded, One entity on several sheets, Reading and the guide as a map

One entity on several sheets

A node can stand on several sheets. It is not a copy: the name, text and attributes are shared, and each sheet has its own position. “+N” on the card says how many more sheets it stands on.

  • Node properties → + another sheet… → a sheet: the node takes a free spot there.
  • In a spread, drag the card into the half of another sheet.
  • Select the card, Ctrl+C, open another sheet, Ctrl+V. On a sheet where the card already stands, Ctrl+V makes a copy — a node of its own; Ctrl+Shift+V makes a separate copy anywhere.
“+N”
In the card's upper right corner. Its tooltip names those sheets; a click opens the list: go to a sheet with the node selected, or Open all as a spread.
The gold mark
The selected entity is marked in gold wherever it shows: in the other half of a spread, in the stack, on the passe-partout. The gold Same ID line joins its appearances.
Stands on sheets
The list in the node's properties: ★ is the primary sheet (the first), ⌕ a smart sheet. Click a sheet to open that appearance; × removes the node from it (the last ordinary sheet cannot be removed).

The project check suggests candidates: when a node has four or more neighbours on another sheet, it probably belongs there too.

See also: Spreads, Project check

Back to sections

Passe-partout: neighbouring sheets

How a sheet shows its links to other sheets, and how to steer it.

The passe-partout: unfolded and folded

Around the sheet is the passe-partout: the nodes of other sheets linked to this one, a group per neighbouring sheet. You see what the sheet is linked to without leaving it.

Folded
The default. Thin strips at the window's edges, a coloured segment per neighbouring sheet. A click on a segment shows its cards over the canvas.
Unfolded
The cards of neighbouring sheets around the sheet: open groups, plates and “More”.

The state is chosen in Passe-partout ▾ in the sheet's upper left corner or in Settings → Passe-partout; the browser remembers it. The sheet pans and zooms inside the passe-partout while it stays put. Unfolded, it always leaves the sheet at least 60 % of the window's width and height.

The unfolded passe-partout: an open group, plates and “More” — the sheet keeps at least 60 % of the window.
The unfolded passe-partout: an open group, plates and “More” — the sheet keeps at least 60 % of the window.

Unfolded: groups

Open group
A header with the sheet's name and node cards — up to six; the rest scroll with the wheel, and the ↑ ↓ counter shows how many are hidden.
Plate
A folded group: the sheet's name and its number of nodes. A dashed outline means the passe-partout folded it for lack of room; a solid one means the choice was made by hand.
“N more sheets”
A hatched plate for sheets without a place. The passe-partout holds up to 16 sheets: first those chosen by hand, then the most connected.
Sides
A group stands on the side where its neighbours are on the sheet: incoming relations lean left and up, outgoing ones right and down. A side holds at most two rows.

Lines to the passe-partout

Every visible card gets its own line. Lines to plates and to scrolled-away cards thin out in steps: up to 80, one from each node; beyond that, one per plate from its main node, with all of them on hover and for the selected node; with very many, only for the selected node.

Moving groups

Drag a group by its header, a plate by the plate itself: on the left and right it slides up and down, on the top and bottom sideways, and next to another side it moves there. The other groups keep their sides. ⟲ in an open group's header gives it back its automatic place and state. A click on a card opens the node itself on its sheet (see Jumping to a node); while linking, it connects to it.

A group slides along its side and moves to another; the sheet pans inside the passe-partout.
A group slides along its side and moves to another; the sheet pans inside the passe-partout.

A selected entity that also stands on a neighbouring sheet is marked in gold on the passe-partout.

See also: Open, fold, pin a group, The Passe-partout ▾ menu and the Sheets filter

Open, fold, pin a group

The unfolded passe-partout decides by itself what to open and what to put into “More”, but you can change any decision — and it is saved in the file.

Open a plate
Click the plate. If the group fits without moving the others (automatic ones may fold, at most three may go to “More”), it opens in place and a note names those left without room. Otherwise it opens over the sheet: Open on the passe-partout keeps it open there, ✕ closes it.
Fold a group
▾ in the header. A group folded by hand stays a plate until you open it.
“N more sheets”
A click opens the list of neighbouring sheets on the right: In “More” on top, On the passe-partout below. Click a row to show that sheet's cards right in the list.
“to the passe-partout” and “to “More””
Buttons on the list rows. “to the passe-partout” puts the sheet on it as a plate; without room, the least connected sheet gives up its place — and is the first to come back when there is room again. “to “More”” takes a sheet off the passe-partout.
The list on the right: sheets in “More” on top, those on the passe-partout below; the buttons move a sheet between them.
The list on the right: sheets in “More” on top, those on the passe-partout below; the buttons move a sheet between them.

The passe-partout remembers itself

Nothing moves without an action of yours. A new neighbouring sheet, a resized window or an opened panel does not throw groups to other sides; a sheet sent to “More” remembers its side. Only Fit, Optimize and Reset manual lay the passe-partout out afresh.

Your choices — a group's place, open, folded, in “More” — are kept in the sheet and survive reloads and file transfers. The list on the right and group scrolling exist on screen only.

If groups opened by hand no longer fit (say, the window got small), the neighbouring sheets are listed on the right, and a note explains why.

See also: The passe-partout: unfolded and folded, The Passe-partout ▾ menu and the Sheets filter

The Passe-partout ▾ menu and the Sheets filter

Passe-partout ▾ in the sheet's upper left corner (and in each half of a spread) steers the passe-partout: its state, which sheets it holds, what to fold and what to reset.

Unfolded · Folded
The two states; the same as Settings → Passe-partout. The number on the button is how many links lead to neighbouring sheets.
Sheets on the passe-partout
Checkboxes for the neighbouring sheets. An unchecked sheet leaves the passe-partout of all sheets and frees its room; all and none switch them at once.
Fold all groups
Folds the groups into plates and keeps them folded.
Reset manual
Forgets the places and states chosen by hand on this sheet and lays the passe-partout out afresh. Undo brings them back.
As a list
The neighbouring sheets as a list instead of the passe-partout; Around the sheet turns it back.
The Passe-partout ▾ menu: its state, which sheets it holds, fold, reset, show as a list.
The Passe-partout ▾ menu: its state, which sheets it holds, fold, reset, show as a list.

The Sheets filter

The same switches are in the Filter panel under Sheets. “and on the overview” hides the unchecked sheets in All sheets too. Nodes on the sheets themselves stay where they are, and a saved view keeps these switches.

In a spread the menu belongs to its half: each has its own neighbours.

See also: Search and filter, Saved views and smart sheets

Back to sections

Views

All sheets, spreads, the stack, contents, architecture, All-to-1, the story and reading — and when to use which.

All sheets

Every sheet of the project on one canvas, like a map of the area: you see where things are and how the sheets connect.

  • Drag a sheet by its header. A node inside a sheet can be dragged too — that changes its real position on the sheet.
  • A double-click on a sheet or Enter opens it.
  • The ⌃ chevron folds a sheet to its header, ⌄ unfolds it; Unfold sheets ▾ among the canvas buttons opens or folds every sheet at once. The folded state is saved in the project.
  • Links: Between sheets and Same ID turn on the links between sheets and the gold lines of shared entities (only the links are on at first: gold lines for the whole map at once are too many; the selected entity always shows its own); Selected only: with a node selected, its links remain; with a sheet selected (click its header or empty space), every link of its nodes.
  • Optimize among the canvas buttons packs the sheets closer: frames fit their content, connected sheets side by side. Undo brings the old layout back.
All sheets: one sheet folded, only the selected node's links shown.
All sheets: one sheet folded, only the selected node's links shown.

Arrow keys move the selected sheet or node by 10 points (Shift: 50), + and − zoom.

When “and on the overview” is checked in the Sheets filter, the unchecked sheets are left out here.

See also: Spreads, Contents

Spreads

Two to six sheets side by side, in a grid or one under another. Lines between the halves show relations and shared entities.

Layout
The ‹ › switch in the views row pages through the layouts: two side by side, two stacked, one tall and two beside it, three, four, five, six. The border between halves can be dragged; a double-click on it splits evenly.
Half
Its header has the sheet list (picking a sheet already open swaps the halves), the node count, ⤢ to open the sheet alone. A tab in the sheet bar puts its sheet into the active half.
Links
Between sheets for relations between the halves, Same ID for the gold lines of one entity, Selected only — with a node selected, its relations and appearances; with nothing selected, the relations of the active half.
A spread, Selected only: the selected node's relations; the gold line is the same entity on the other sheet.
A spread, Selected only: the selected node's relations; the gold line is the same entity on the other sheet.

Drag a card into the other half and the node stands on that sheet too. Each half has its own passe-partout; nodes already visible in the other half are left off it — a line goes straight to them. Sheets without a manual layout arrange themselves until you move a node.

On a phone the halves stand one under another and scroll down; there are no lines between them there, so each half's passe-partout shows all its neighbours.

See also: One entity on several sheets, Which view to use

Stack

Sheets lie as plates one above another, so you see one entity pass through them. Good for a slide or an article.

A plate is a sheet in miniature: its cards lie flat on it, in rows, where they stand on the sheet, and the sheet's name is written on its front edge. One stack holds up to five sheets; more make several stacks side by side. Cards lie with gaps between them and never on one another; a crowded plate shows the nodes linked to other plates, and all the plates of the stack grow with the most crowded one. Sheets and stacks ▾ decides which sheets lie in the stacks and how many stacks there are: a tick per sheet, Stacks − N +, and each sheet's own stack. The tabs in the sheet bar also put a sheet in or take it out.

The stack: sheets as plates; gold lines — entities on several sheets — run to the plate's edge and pass to the other plates.
The stack: sheets as plates; gold lines — entities on several sheets — run to the plate's edge and pass to the other plates.

Lines between plates go as on a multi-layer board: along the plate from the card to its right or front edge, through a via there, straight down or up to the same point on the other plate and along it to the card. A gold line joins one and the same entity, a dashed one a relation between nodes of different sheets. Between stacks only the selected node's lines are shown.

Sheets and stacks ▾
Which sheets are in the stacks (all, none or tick by tick), how many stacks (up to five sheets in each; the sheets are shared out in order) and which stack each sheet is in.
Look ▾
Tilt (how high above the plates you look), the spacing between them, labels (all, or the selected node and its neighbours) and how to split into stacks: in order, by sheet type, by a tag, or by hand — drag a sheet tab onto a stack or press ⇄ by a plate.
Links
Between sheets for relations between plates, Same ID for the gold lines of one entity, Selected only.
Align
A canvas button: shared nodes take one point on every plate, so an entity shows through the stack as one column. As on the sheets turns it back. The map does not change: the stack is only a view.
As a picture ↓
SVG for an article, PNG for a slide; the buttons stay out of the picture.

On a crowded plate you see the nodes linked to other plates and the selected one; the rest are counted — “N more” on the right edge. A click on a node selects it, a double-click opens it on its sheet; a click on a plate's edge makes it active, a double-click opens the sheet. ↑ and ↓ choose the active plate.

See also: Which view to use, One entity on several sheets

Contents

How the project is built: the sheets and the links between them, as a diagram or a list.

Diagram
Sheet cards and, between any two linked sheets, one quiet line with a card of numbers: “→ 11 · ← 3 · ◆ 2” — relations each way and shared nodes. Point at a card of numbers to light both sheets and their line; a click on it opens the two sheets side by side. A click on a sheet highlights its links, a double-click opens it, dragging moves the card; places are kept in the project. Diagram · List is above the diagram. Optimize among the canvas buttons brings connected sheets closer. The diagram scrolls and zooms the same way as a sheet.
List
Sheets as sheets of paper with the cards of their nodes. A sheet unfolds to its nodes, a card to its text and attributes. Cards unfolded in the list are unfolded on the sheets too. Unfold all ▾ unfolds every sheet or every card. Below the sheets, pairs of linked sheets with a “bridge”: which nodes join them.
Lines
One per pair of sheets, faint: the directions of the relations and the shared nodes are told by the numbers on the pair's card, not by the line.

See also: All sheets, Architecture

Architecture

Sheets as blocks in the manner of C4: a name, a type, a description, and inside them the nodes you choose. A picture of the system made of sheets and its key parts.

  • Sheets: in the views row chooses which sheets to show; the tabs in the sheet bar do the same.
  • Inside sheets: — which node types stand inside the blocks of their sheets. A single node is added or taken away with the + on a block. The choice is kept in the file.
  • Relations are not drawn until you turn them on, so the picture stays clean.
  • Blocks can be dragged; Optimize among the canvas buttons packs them tight, in rows in the proportions of the screen, connected sheets side by side. One step undoes it. The picture scrolls and zooms the same way as a sheet.
  • Unfold all ▾ among the canvas buttons shows the start of the text (Text), the attributes (Attributes), the picture (Pictures) or all of them on the nodes inside the blocks; Fold all goes back to the name and [type]. The browser remembers the choice. When you zoom out, names keep their size on screen — about 12 pixels for nodes and 15 for blocks — and show in full, taking the room of the small text under them. The rest stays small and turns paler, and the [type] and the node count disappear when they can no longer be read.
  • A click on a block shows the sheet's properties, a double-click opens it; As a picture gives SVG or PNG.

See also: Contents, Which view to use

All-to-1

Every entity once, on one field, in its own arrangement: the whole picture.

The arrangement here is separate and does not touch the sheets; text and attributes are shared. Dashed lines join nodes that share no sheet. Optimize works here as well. New nodes are not created in this view, and there is no passe-partout.

See also: Which view to use

Story

A story is the steps by which you lead a reader through the map. A step is a view of the map with a title and a few lines: press Watch and the screen moves from step to step by itself.

  1. Open the view you want: the view mode, the sheets, the zoom, the filter; select and unfold the cards you need.
  2. Story → Add a step: the step remembers all of it and the state of the passe-partout.
  3. Story → Steps…: the story's title, each step's title and text; the order changes by dragging or ↑ ↓; Show opens a step, × deletes it.
  4. Story → Watch — from the first step.

A bar at the bottom of the screen: “Step 3 of 12”, the title, the text and arrows. → or Space goes on, ← back, Home and End to the first and the last, Esc ends it. When a step is on the same sheet, the camera glides to its new place. After the story the passe-partout returns to your choice.

The story is kept in the map file. A link with ?story=1 opens the map on its first step, and a copy without the private part drops the steps on private sheets. The example PlyLoom about itself in Project → Examples is a map with a twelve-step story.

See also: Reading and the guide as a map, Saved views and smart sheets

Reading and the guide as a map

Reading shows the same map without editing — to walk someone through it without moving anything by accident.

  • The Reading button in the first row, or a link with ?view=1. Back to editing: Edit in the bar at the bottom left.
  • No panels, no +, no dragging, renaming or editing; the menus are Settings, Filter, Story and Help.
  • A click on a card unfolds it whole. A click on a relation leads to the other end — on its own sheet if it stands on another; ← Back returns to where you were.

The guide as a map

Help → The guide as a map opens this guide as a PlyLoom map: parts are sheets, articles are nodes, an article's text is the node's text, see-also links are relations. It opens read only and apart from your map: yours stays as it was and is not written over. ← Back to your map is in the bar at the bottom.

See also: Story, Jumping to a node

Which view to use

The views show the same data; choose by the question you are working on.

TaskView
Work on one side of the topicSheet
See the whole project and the links between sheetsAll sheets
Compare two or three sheets and connect their nodesSpreads
Show an entity passing through sheets — on a slideStack
Understand how the sheets relateContents
Draw the system of sheets and its main partsArchitecture
See every entity at once, without sheetsAll-to-1
Walk someone through the mapStory and Reading

Back to sections

Dictionaries, attributes and search

The map's shared lists, node fields and finding what you need.

The map's dictionaries

Dictionaries are the shared lists of the whole map: what kinds of sheets, nodes, relations and card fields exist. Change them here and they change everywhere. Edits are a draft until you press Save types.

Sheets
Sheet tags, sheet types and the project's sheets: name, type, tags, Delete empty sheet.
Nodes
Node types for the palette: name (and an English name), description — the palette tooltip, colour, notation, PlyLoom base type and node shape.
Relations
Relation types: name and description.
Attributes
Node fields: name, description and data type.
Dictionaries → Attributes: a node field, its data type and options.
Dictionaries → Attributes: a node field, its data type and options.
  • A type's shape applies to new nodes; existing ones change when you change their type.
  • Built-in types can be edited but not deleted. A type in use — by nodes, relations, sheets, a view or a smart sheet — cannot be deleted; the Used counter shows why.
  • An attribute's data type is locked while it is in use. A choice option that nodes use cannot be deleted.
  • The English name is shown when the interface is in English.

See also: Node attributes, Notations and shapes

Node attributes

An attribute is a node field with a value: Status, Due date, Owner, Score. Attributes drive the filter and dividing sheets.

Data typeValue
Textup to 10,000 characters
Numberany finite number
Yes / noyes or no
Datea calendar day
URLa full http or https address
Choiceone of 1–100 options
Multiple choiceseveral options
  1. Create the attribute: Dictionaries → Attributes → + Create attribute, choose the data type and save the types.
  2. Unfold the card and click its attributes, or open the node in Properties, and press + attribute.
  3. Fill in the value. An invalid value is not saved.

× removes an attribute from the node. An attribute without a value (Not set) is information too: the filter can find such nodes.

The lock 🔒 next to a value makes only that value of this node private: on other nodes the attribute stays ordinary. An attribute private on every node is marked in the dictionary — then the lock shows at each of its values and cannot be lifted here.

Sheet properties → Lay the sheet out by a key → an attribute gives one sheet per value.

See also: Search and filter, Sheets: create, set up, divide

Search and filter

The filter selects nodes by text, types and attributes, in every view at once. The file does not change: the filter acts on the screen only.

Search

The search line looks for the text, ignoring case, in the checked areas: name, body, tags, sheet (a sheet's name or tag), attributes. With several words, all must be found, each in any area. Enter goes to the next match; recent queries are offered in a list.

Conditions

  • Node types and Relation types — checkboxes with counts; Select all and Turn off all.
  • Attributes — + Attribute condition: a range for numbers and dates, “contains” for text and URLs, options for yes/no and choices, “any of” or “all selected” for multiple choice. Every condition also has “empty” and “no attribute”.
  • Sheets — which neighbouring sheets the passe-partout shows.
  • Different conditions apply together (AND); checks inside one condition mean any of them (OR).
The filter panel: search, sheets on the passe-partout, types and attribute conditions.
The filter panel: search, sheets on the passe-partout, types and attribute conditions.
Hide
Filtered objects are not drawn — fast on large maps.
Dim
Filtered objects fade in place and stay clickable.

A node you create, open or edit does not vanish when it does not match — until the filter changes. Sheet tabs with matches are underlined and the rest fade; Matches only keeps the tabs with matches. The folded panel shows a status line: how many nodes are shown, how many relations hidden, Adjust and Show all.

See also: Saved views and smart sheets, The Passe-partout ▾ menu and the Sheets filter

Saved views and smart sheets

A view remembers the filter, the view mode and the set of sheets. A smart sheet gathers nodes by the filter's conditions by itself.

Save a view
Filter → Saved views → a name → Save view. The view travels inside the .plyloom.
Actions
Open, Rename, Replace with current, Delete.
Smart sheet
Also there: Create smart sheet — a sheet holding every node of the ordinary sheets that matches the node conditions (relation types and passe-partout sheets do not count). Its membership updates itself on every edit.
  • New nodes are not added to a smart sheet — the filter decides its membership.
  • It cannot be laid out by a key, and a node cannot be removed from it by hand.
  • Its query changes with Replace the query of “…” with the current filter in Saved views.
  • Sheet properties → Make an ordinary sheet turns it into an ordinary one; the nodes stay.

See also: Search and filter

Back to sections

Project structure

Sheets and their boundaries, the check and how cards look.

Sheets: create, set up, divide

A good sheet answers one question. A new sheet is + in the sheet bar; set it up in the properties.

+ in the sheet bar
A new sheet waits for its name right in its tab; type, tags and colour are in the properties. A double-click on a tab renames the sheet.
Sheet properties
Name, type, tags, colour and description; how many nodes there are and how many are shared with other sheets; the sheet's nodes; links to other sheets — relations and shared nodes with each.
Delete sheet
Only an empty sheet, and only while at least one ordinary sheet remains.

Lay a sheet out by a key

When a sheet answers several questions, divide it: Sheet properties → Lay the sheet out by a key → node type, node tags or an attribute → Lay out. You get one sheet per value (“Sheet · value”); nodes with several values stand on several sheets, nodes without a value on a No value sheet. Relations and shared entities are kept; Undo restores the previous map.

Sheets can also be renamed and deleted in Dictionaries → Sheets, together with sheet tags and sheet types.

See also: How a map is built, Project check

Project check

Help → Check project looks for gaps and suspicious links. These are hints, not verdicts: you decide what a sheet boundary means.

HintWhat to do
✖ An edge refers to a missing nodeA broken reference in the file: delete the relation.
▲ A sheet has no connectionsNo links to other sheets and no shared nodes: connect it, or ask whether it is needed.
▲ A node has no connectionsA loner: connect it or delete it.
▲ Listed on a sheet but not connected thereAn appearance without neighbours: remove the node from the sheet or connect it.
▲ N neighbours on sheet “…” — it probably belongs thereA candidate for + another sheet….
· A sheet is empty, or much larger than the othersFill it or delete it; read a large one with the filter.
· A hub, no body, a name that is too longA hub has very many relations: check whether it is a topic. Names over 44 characters: shorten them and move the text into the body.
· Stands on many sheets; a contradictionA node on half the sheets may be a topic. A “contradicts” relation marks a disputed point.

Click a row to go to the node or sheet; copy report copies the list as text.

See also: Sheets: create, set up, divide, Relations

Settings

Settings change only the view on this screen — never the map or its files. Four parts: cards, the passe-partout, the screen and the counts.

Cards
Shape by node type: turn it off for cards of one shape with a strip in the sheet's colour; notation shapes keep their geometry. Pictures on folded cards: the small pictures under the name.
Passe-partout
Unfolded or Folded — the same as Passe-partout ▾ on the canvas.
Screen
Menus and views above the map: without them only the sheet bar stays, and the view is chosen there. Ctrl+. toggles it. PlyLoom mark in the canvas corner: a barely visible label at the lower left, so a screenshot shows where the map was made; it can be turned off.
Counts in the sheet bar
How many sheets, nodes and filled attribute values the map has: “6 (1) · 22 (2) · 26 (1)”. In brackets, the private part: how much stays only with you and does not go into a copy for sharing. A node counts once, however many sheets it stands on. Next to them, the size of the .plyloom file and, if the map has pictures, how much of it they take: “3.2 MB ▧ 2.5 MB”. A click on the numbers opens Settings.
Counts in the file name
The same numbers at the end of a downloaded file's name: renovating-the-flat-2026-10-11_S6(1)E22(2)A26(1).plyloom — S sheets, E nodes (entities), A attribute values, the private part in brackets; with nothing private, there are no brackets. Copies of a map can be compared right in the folder, without opening them. Versions from the history are named the same way, and a copy for sharing carries the numbers without the private part: …-public_S5E20A25.plyloom.
Summary
Under the counts: Sheets · Nodes · Relations · Attributes · Pictures and the size of the map, text and pictures apart.

The browser remembers these settings. To hide or dim nodes and relations by type, tag and attribute, use the Filter.

Notations and shapes

A node can look like a PlyLoom card, a flowchart block, a BPMN element or a Canvas card — its type decides.

  • Type shapes: rectangle, rounded, ellipse, diamond, parallelogram, cylinder, document, text without an outline, and group — a backing frame drawn under other nodes.
  • BPMN: events are circles with start, end and intermediate rings; gateways are diamonds with ×, + or ○; labels sit below the shape.
  • The palette groups Flowchart, BPMN and Canvas hold ready-made types; your own types appear under Project types.
  • A sheet's notation (PlyLoom, draw.io, Obsidian Canvas, BPMN) tells where the sheet came from: imports set it.

The example One payment, six models in Getting started shows one topic in six notations joined by traceability.

See also: The map's dictionaries, Import

Back to sections

Files and exchange

Where the map lives, how to keep it, bring it in and hand it over.

Saving and backups

The map is saved in the browser after every edit, and before large changes the browser puts its previous version aside by itself. But all of this lives in one browser: a .plyloom file is the copy that survives clearing the browser and a new computer.

Autosave
Every edit goes into the browser's storage within a second, and at once when the tab closes or is hidden. The state is on the right of the toolbar: saving…, saved in browser, saved in browser and file. Click it to open the version history.
Version history
Project → Version history…. PlyLoom keeps the map by itself before large changes — deleting several nodes or relations, opening another map, an example, a new project, edits from an AI, Divide the sheet — and every 10 minutes of work. The last 20 versions stay, plus one for each day of the past month. Any of them can be restored or downloaded; restoring can be undone too, and the open map goes into the history first.
Several tabs
One map can be open in two tabs or two copies of PlyLoom.html: they know about each other. A tab with nothing unsaved shows the fresh version by itself. If both have edits, nothing is overwritten: a bar appears and you choose which version to keep. The other goes into the version history.
Copy as a file
Project → Download .plyloom: a file with the whole map, named after the project and the date, with the number of sheets, nodes and attributes at the end, the private part in brackets (_S6(1)E22(2)A26(1), see Settings). This is the backup: keep it like any document, ideally in a folder with its own version history (Google Drive, Dropbox, OneDrive). When there has been no copy for a long time, a quiet reminder appears under the toolbar.
Save to a file
In Chrome, Edge and other Chromium browsers on a computer: Project → Save to a file…. Pick a file once, and from then on every edit is written to it too, as in an ordinary program. After a restart the browser may ask you to confirm access to the file with one click. If another program changed the file, PlyLoom does not overwrite it until you agree.
Open
Project → Open .plyloom…: a check and a preview — how many sheets, nodes, relations and attributes the file has and how many the open map has now — then the current project is replaced. The previous map goes into the version history, and Undo brings it back too.
Undo and redo
100 steps until the page reloads. Quick edits in a row merge into one step. If you undo several steps and then make a new edit, the undone work is not lost: it is kept in the version history.
Version
A downloaded file remembers the fingerprint of its version — 12 characters computed from the content — and the fingerprints of the versions it came from. Fingerprints only, no earlier content. They tell PlyLoom how two copies are related.

If the browser does not let the page save — a private window, site data blocked, a full disk — a red bar appears at the top with a Download a copy button. Do not close the tab before you download the file.

A browser may delete site data when the disk runs low; Safari erases the data of a site not opened for a week. So once there is work in the map, PlyLoom asks the browser to keep it for good; Firefox asks for your permission. The browser's answer is shown in the version history.

Examples from the website open in separate storage and do not touch your map. If the previous save cannot be read, a banner offers to download it — keep that file before you go on.

See also: Import, Export, Edits item by item: comparing and AI answers

Import

PlyLoom opens its own format and adds sheets from draw.io, Obsidian Canvas and BPMN. Every import starts with a preview.

FormatWhat happensWhat comes along
.plyloom, .jsonreplaces the projecteverything; older versions are read too
draw.io (.drawio, .xml)adds a sheet per pagepositions, sizes, basic shapes, colours, font, rotation, bend points; no images or attachments
Obsidian Canvas (.canvas)adds a sheet per filetext, links, files and groups, positions and colours; no file contents
BPMN (.bpmn, .xml with DI)adds a sheet per diagramtasks, events, gateways, flows, pools and lanes; documentation becomes the body
CSVreplaces the projectnodes and relations from two files
Before opening, a preview: how many sheets, nodes and relations will come.
Before opening, a preview: how many sheets, nodes and relations will come.

CSV

Pick both files at once: nodes.csv with the columns id, name, kind (the type; empty means Entity), body, sheets (sheets separated by |); edges.csv with from, to and the optional id, kind, label. Nodes are laid out on a grid.

  • Up to 128 MiB at a time; above 20 MiB PlyLoom asks first.
  • An error rejects the whole file: the current map stays, and the reasons appear in Project check.
  • A BPMN element shown in several diagrams becomes one entity on several sheets.

See also: Export, Working with AI: a prompt out, edits back

Export

Keep the .plyloom to work without losses; the other formats are for exchange.

.plyloom
The whole project: sheets, dictionaries, views and layout.
.drawio
Each sheet becomes a page of editable shapes; links to other sheets become external cards. Saved views, smart queries, sheet descriptions and the overview and All-to-1 layouts do not come along.
Canvas (Obsidian)
A sheet becomes a group, each appearance a card with its ID. Types, attributes and shapes do not come along.
JSON v1 tree
An old format without dictionaries; unavailable for projects with types.

See also: Saving and backups, Assemble: a document, a table, a diagram

Assemble: a document, a table, a diagram

The map is the source, a file is a build. Project → Assemble a document… makes a document, a table or a diagram from the map, with no AI and no network. Change an entity once and assemble again: every file that holds it changes.

Document
Four formats, by where the document goes. Word (.docx) opens in Word, Google Docs, LibreOffice and Pages: headings in the navigation pane, a linked table of contents, page numbers; from there it is an ordinary document to edit. PDF — PlyLoom opens the browser's print dialog, where you choose Save as PDF. HTML is one page that works without internet in any browser. Markdown is for wikis, Git and Obsidian. A sheet becomes a section, a node a subsection with its text, attributes and relations. The order is how the sheet reads: top to bottom, left to right; a number attribute with the ID order overrides it. The header and the footer name the map version the document was built from.
A shared node
A node on several assembled sheets is given in full once, then as a “see the section” reference. The box “in full in every section” repeats it whole: that is how a CV per vacancy or a checklist per procedure is assembled.
Table
CSV: nodes with attributes, relations, or a rows × columns matrix. A matrix cell holds the relations between its row and column; the last column counts related columns, and zero is an uncovered row: a requirement without a check, a role without a process.
Mermaid diagram
A sheet as code for a mermaid block in GitHub, GitLab, Obsidian, Notion and wikis. Relations to other sheets end at double-framed cards.

What to assemble: the current sheet, every sheet, the chosen ones, or a saved view — then only the nodes its filter lets through go in. A smart sheet by status gives a status report in one click.

Private content stays out of an assembly until you tick it in: private sheets and nodes with one box, private attributes and locked values with another. That way a CV gets your contacts but not your own notes.

Saved assemblies

At the bottom of the window, Save these settings as an assembly keeps the sheets, the format and the boxes in the map file under your name for it, say Memo for the manager. Assembling again is then one button: Project → Assemble again: … or the named button at the top of the window. The pencil next to it opens its settings, the cross deletes it; the map itself does not change.

The What goes out button shows the file before it is downloaded: what goes into it, what stays with you, and the text itself with a search.

See also: Private: what never leaves the map, Saved views and smart sheets, Export

Private: what never leaves the map

A sheet, a node, an attribute and even one value of an attribute on one node can be marked private. Private content never goes to an AI, and goes into a file for someone else only if you decide so.

A private sheet
The box in the sheet's properties. Nodes that stand only on it become private. A node that also stands on an ordinary sheet stays visible there: its text is shared by all its sheets.
A private node
The box in the node's properties: private on any sheet. Its card shows a lock.
A private attribute
The box in the attribute dictionary (Dictionaries → Attributes): its values on every node show to you but stay in the map. For example: contacts, income, internal sources.
A locked value
The lock 🔒 next to a value in the node's properties or the unfolded card: only that value of this node becomes private; the attribute stays ordinary on other nodes. Say, Source is open everywhere except one node whose source is internal. A file for someone else holds neither the value nor a trace of the lock.

Two ways out

  • An AI — through the clipboard. The prompt is text you copy yourself. Private content is cut out of it with its relations, and a private sheet cannot go out as the part. If an AI returns the whole project, the comparison does not offer to delete your private content.
  • People — through files. Project → Download .plyloom asks who the file is for when the project has private content: “For me — the whole file” or “For someone else — without private content”. The copy without it is marked, and when it comes back edited, Compare with a file does not offer to delete what it never held.
  • Documents, tables, draw.io are built without private content until you tick it in. Canvas and the JSON tree never take it.

What goes out

Before the map leaves, you can see it as it will go: in the who-is-this-file-for question, in the AI window before the prompt is copied, and in the assembly window. Left, the sheets and nodes: what goes, what stays with you, what goes without the values of private attributes, what is not part of this output. Right, the very text that goes, with a search that also says how many times the same words stand in the whole map.

The output check looks for each private name, line of text and value in that text. A match is not always a leak: the same words can stand in an ordinary node, and a value copied by hand into an ordinary node goes with it. So the check shows where the private string comes from and where it was found. It does not look for short strings and numbers: they would match ordinary text.

Privacy is your own mark: it never arrives with edits from a colleague or an AI and is not compared. AI edits that would remove or overwrite private content are skipped — the AI never saw them. A backup before replacing the project is always whole.

See also: Working with AI: a prompt out, edits back, Assemble: a document, a table, a diagram, Saving and backups

Working with AI: a prompt out, edits back

PlyLoom does not connect to an AI and sends nothing. The AI menu opens one window for the whole exchange: the prompt is copied into any chat, and the answer is pasted into the same window and checked before anything changes.

Refine the map
The map — whole, or a sheet with its neighbours — goes into the prompt with the task; the answer is a list of edits. You tick what you want and apply it; the edits stay marked on the map until you accept or reject them.
A report, a letter, a checklist
These are tasks of Refine this map: the text comes back as a new node (Report, Questions and answers, Instructions, Cover letter, Checklist) linked to the nodes it relies on. Project → Assemble a document turns it into Word, PDF, HTML or Markdown.
Where the facts come from
For a new map: Only from my materials — what the materials do not say is not on the map, and without materials the model will not fill it from memory; A draft from general knowledge — anything unchecked gets the source “general knowledge, unchecked” and a confidence of 30 at most; Mixed — materials first, gaps marked.
A new map from materials
A prompt without the map: you add your texts and files in the chat, and the answer is a whole map. A preview, then Open and replace. If another map is open, the window offers to download it first or to start from an empty project.
Base and task
A prompt is a base plus a task. The base is the exchange contract, by what comes back. The task is the text at the end: a ready-made one (Answer open questions, Check doubtful claims, A report from the map…) or your own — you see it and can edit it. Show the prompt text shows the whole prompt.
Question and answers
Present in every map: the node type Question in the palette and the relation answers. A question no such relation points to is open. Source and confidence: the ready-made tasks ask the model to keep them in the attributes source and confidence, and to define them if the dictionary lacks them.
The answer
Paste the whole answer into the field at the bottom of the window — text around the JSON and code fences do no harm — and press Show the edits or Show the map. Closed the window? AI → Paste an AI answer…. An answer saved as a file opens with The answer is a file….
The AI window: two tabs at the top and three steps — the task, the materials or what goes out, the prompt and the answer.
The AI window: two tabs at the top and three steps — the task, the materials or what goes out, the prompt and the answer.

The window opens on the New map or the Refine the map tab. Each has three steps. A new map: the task → where the facts come from → copy the prompt and paste the answer. Refining: what goes out (the whole map or a sheet with its neighbours) → the task → the prompt and the answer.

Private content stays out of the prompt: private sheets and nodes go with their relations, and so do the values of private attributes and locked values. What goes into the prompt shows the map as the AI will see it and searches the prompt text itself.

The prompt asks the AI not to invent facts, to use null for unknowns, to give each sheet one question and to preserve your project. After a new map, review the facts and sheet boundaries; Check project finds isolated nodes and empty sheets.

Refine a finished map with edits rather than a new project: Refine the map asks the model for a list of edits only, so it cannot rewrite or lose what it did not touch. With a colleague the exchange goes through files: Project → Download .plyloom and Project → Compare with a file. You decide which service to trust with your materials.

See also: Edits item by item: comparing and AI answers, Private: what never leaves the map, Import

Edits item by item: comparing and AI answers

Edits from others never enter the project by themselves. PlyLoom lists them by sheet, you tick what to accept, and everything is applied as one step you can undo.

Compare with a file
Project → Compare with a file… — a copy from a colleague, or a file from an AI. Differences are found by meaning: nodes, texts, attributes, relations, sheets and dictionaries. Moved cards are not listed unless you ask to bring positions too.
Compare with current
The preview that opens a .plyloom offers to compare instead of replacing the project.
An AI answer
Pasted in the AI window the prompt came from, or with AI → Paste an AI answer… — a list of edits, or the whole project, straight from a chat; text around the JSON does no harm.
Edits are grouped by sheet. The conflict is not ticked: you already have another value there.
Edits are grouped by sheet. The conflict is not ticked: you already have another value there.

Reading the list

  • + adds, ~ changes, − removes. A changed text is shown word by word: what goes is struck through, what comes is highlighted.
  • Conflict — the edit expects one value and you already have another. Such items start unticked; tick one to replace anyway.
  • An item that needs another (a relation to a new node, a node on a new sheet) does not apply without it; the number of the needed item is shown.
  • Error — the edit does not fit the project: no such sheet, an unknown type, a text that is too long. It cannot be ticked.

Unticked items are rejected when you apply. To decide later, use Later (or the cross): the list waits in the bar at the bottom of the screen — “Not reviewed: N · open the list” — and comes back with the same ticks. Back to the edits in the AI menu (AI edits) or the Project menu (edits from a file) does the same. Reject all, in the list or in the bar, drops the edits without applying anything.

After applying: accept or reject

Applied edits are already in the project but stay marked, like tracked changes in a text editor: accept or reject them — all at once in the bar at the bottom of the screen, or node by node.

At the bottom, the bar of these edits: colours, removed, ‹ ›, Accept all, Reject all. On the right, the selected changed node: what was done to it, Accept and Reject.
At the bottom, the bar of these edits: colours, removed, ‹ ›, Accept all, Reject all. On the right, the selected changed node: what was done to it, Accept and Reject.
  • Outlines show where the edits went: green for a new node, amber for a changed one — its name, text, type, an attribute, its sheets or the relations at its ends. The nodes may be on any sheet: ‹ › in the bar visit them one by one and switch sheets.
  • Select a marked node and the properties panel shows what was done to it, with two buttons: Accept keeps it and clears the mark; Reject returns this node as it was — its fields, sheets and the relations at its ends; a new node is removed.
  • Accept all in the bar keeps every edit and clears the marks. Reject all takes back everything the edits did, removed items included. Anything you changed yourself after applying stays.
  • What was removed is not on the map; “removed: N” in the bar lists the removed nodes, relations and sheets. Dim the rest leaves only the marked nodes in view.

Nothing unreviewed gets lost: a list put aside and applied edits not yet accepted are kept in the project itself. They survive a reload and go into a downloaded .plyloom, so the review can continue later or on another computer. So that edits can still be rejected later, the project keeps the old values of the places they touched; after Accept all or Reject all that data leaves the project.

Removals

  • Remove node deletes it from every sheet together with its relations; the item says how many.
  • Leaves sheet takes the node off one sheet only; it stays on the others. Its last sheet cannot go this way: that is an error suggesting to delete the node itself.
  • Attribute removed means the node no longer has it; empty means the attribute stays without a value.
  • Without a common version, removals in a comparison start unticked: they may be your new nodes. If you changed a node that the edits remove, that is a conflict.

Copies that went separate ways

From version fingerprints PlyLoom tells what has arrived: a continuation of your version, an older copy, or a copy edited alongside yours. The last case needs the common version. Its content is kept only in this browser — the last eight versions you opened or downloaded — and goes nowhere.

  • The common version is here: only their edits are listed, and whatever you changed too is marked as a conflict.
  • It is not here: the differences mix both sides, and removals start unticked because they may be your new nodes.

When refining a map with an AI, ask for edits, not the whole project: AI → Refine this map…. The model returns only what it changes and cannot lose the rest.

See also: Working with AI: a prompt out, edits back, Saving and backups

Back to sections

Reference

Keys, phones and the technical description.

Keys and gestures

Everything works with a mouse or fingers; keys speed up frequent actions.

ActionHow
Undo · redoCtrl+Z · Ctrl+Shift+Z or Ctrl+Y (⌘ on a Mac)
Previous / next sheetAlt+← · Alt+→
Delete the selected node or relationDel (⌫ on a Mac)
Copy the selected cardCtrl+C (⌘+C on a Mac)
Paste: on another sheet the same card, on its own sheet a copyCtrl+V
Paste as a separate copyCtrl+Shift+V
Text from another program as a new noteCtrl+V on a sheet
Select a relationclick its line
Clear the selection, close menus, cancel linkingEsc
Next search matchEnter in the search line
Select a card from the keyboardEnter or Space
Pan · zoomdrag empty space or the wheel · Ctrl + wheel, pinch
Scroll horizontallyShift + wheel
New node · relationdouble-click empty space · drag a dot on a card's edge
Unfold · fold a card↘ · ↖ in the lower right corner
Rename a node with the mousea click on the name of a selected or unfolded card
Rename the selected cardstart typing · Enter or F2 to edit
Fold or show the topCtrl+.
Story: a step on and back · end→ · ← · Esc
Overview: move a sheet or node · zoomarrow keys (Shift: faster) · + −
Stack: the active plate↑ · ↓
Paste a picture into the selected nodeCtrl+V

Phones and tablets

On a narrow screen everything is the same; the panels slide in from buttons.

  • The Nodes and Properties panels open from the ☰ and ⓘ buttons in the sheet bar; tap a type to add a node.
  • Pan the sheet with one finger, zoom with two.
  • In a spread the halves stand one under another: scroll down, or pick a sheet from the tabs.
  • Fold the top with ⌃ and the sheet gets almost the whole screen.

See also: Spreads

Technical description

One HTML file without a server: what is inside, where it is stored and how the format works.

Application
One HTML file with the interface, code and illustrations inside. No server, installation or account; it also runs from disk (file://). The page's security policy forbids network requests (connect-src 'none') and external images, and allows one script only — its own, named by its SHA-256: no inline handler or other code on the page runs.
Browsers
Current Chrome, Edge, Firefox and Safari (Chrome/Edge 107+, Firefox 104+, Safari 16+). JavaScript is required.
Storage
The map and its version history live in this browser's IndexedDB. Every write is numbered: a tab writes only if storage still holds the version it continues from — the check and the write share one transaction — and tabs tell each other about every write (BroadcastChannel and the storage event). The version from before a large change is written in the same transaction as the change. If IndexedDB is unavailable, localStorage is used, without version history; if both are blocked, the work stays in memory and a red bar says so. Save to a file uses the File System Access API. Node pictures are kept apart, each once: the map and the versions in the history hold references only, and pictures nothing refers to are deleted in the background. Recent searches, folded panels and palette groups are kept in the browser too.
Identity
A node has one id. Memberships are the sheets array; the first sheet is the primary one. Positions (pos) and geometry (appearance) are per sheet; name, body, type and attributes are shared. Gold lines are not stored — they follow from memberships.
Passe-partout
Its two states — unfolded and folded to strips — are remembered by the browser, apart from the map (the earlier “hidden” reads as folded). Unfolded, it is tied to the window, not to the sheet; the sheet keeps at least 60 % of the width and height, and the passe-partout holds up to 16 sheets. Choices made by hand are stored in sheet.externals under the neighbouring sheet's key: side and along — the side and the place along it (a share of its length), state — open, folded or parked (in “More”). Neighbours moved by an unfolded card move on screen only.
Limits
No node or sheet limit. Import up to 128 MiB, confirmed above 20 MiB. Undo keeps 100 steps until the page reloads; the version history keeps the last 20 versions and one a day for a month, up to 100 MB (but at least three versions). A picture is a PNG, JPEG, WebP or GIF up to 25 MB and 40 Mp; it is kept small: 640 and 192 px on the long side, WebP.
Privacy
The map never leaves your computer. RU/ENG translates the interface and built-in examples; your text and IDs are never rewritten.

File format

.plyloom is JSON in UTF-8, version 3; a map with pictures is version 4 (PlyLoom 0.14 and later open it):

{
  "version": 3,
  "title": "…",
  "types": { "nodes": [], "edges": [], "sheets": [], "tags": [], "attributes": [] },
  "sheets": [{ "id", "name", "typeId", "tags", "color", "description",
               "overviewPos", "overviewSize", "overviewCollapsed", "tocPos", "archPos",
               "externals": { "<sheet id>": { "side", "along", "state" } } }],
  "nodes":  [{ "id", "name", "kind", "body", "attributes",
               "sheets": ["s1", "s2"],
               "pos": { "s1": { "x", "y" }, "s2": { "x", "y" } },
               "appearance": { "s1": { "width", "height", "shape" } },
               "image": { "id": "img_<16 hex>", "alt" } }],
  "edges":  [{ "id", "from", "to", "kind", "label", "directed", "routes" }],
  "savedViews": [], "flatPositions": {},
  "architecture": { "kinds", "nodes", "off", "hiddenSheets", "links" },
  "story": { "title", "steps": [{ "id", "title", "text", "view", "camera", "selected", "expanded", "frame" }] },
  "assets": { "img_<16 hex>": { "type": "image/webp", "w", "h", "micro": "<base64>", "mini": "<base64>" } }
}

Pictures: the id is the start of the SHA-256 of the picture itself, so one picture on several nodes is kept once. A file carries only the pictures nodes refer to; on reading, each is checked — type, size, base64, only the described keys.

Attributes: definitions live in types.attributes (text, number, boolean, date, url, select, multi), values in node.attributes by definition id; null means attached but empty.

See also: Saving and backups, Open, fold, pin a group

Back to sections

Recipes

Ready-made scenarios: how to build a map for a particular job.

Untangle an article or a document

A long text becomes a map: claims, evidence and questions, each on its own sheet.

  1. Create a project with three sheets: Claims, Evidence, Questions. You can ask an AI for a draft — see A map from your materials with AI.
  2. On Claims, one node per claim of the author, of type Hypothesis or Decision; the quote goes into the body.
  3. On Evidence, facts and sources; connect them to the claims with “supports” or “contradicts”.
  4. Anything unclear goes to Questions, with a “reference” relation to the claim.
  5. Open Claims: the frame shows which claims are backed and which hang without evidence. Check project lists the contradictions.

A Confidence attribute (a choice: high, medium, low) and a filter on it quickly leave only the disputed points.

See also: Relations, Working with AI: a prompt out, edits back

Requirement → architecture → test

Traceability: from a requirement to the component and to the test that checks it.

  1. Sheets Requirements, Architecture, Tests.
  2. Requirements use the Requirement type and the attributes Priority and Status.
  3. Components connect to requirements with “implements”, tests with “supports”.
  4. A spread “one tall and two beside it” shows the whole chain; Selected only shows the path of one requirement.
  5. A filter Status: empty finds requirements without a decision, and a smart sheet on Status gathers, say, everything in progress from every sheet.

The built-in Software example (Project → Examples) is built exactly this way.

See also: Spreads, Saved views and smart sheets

Research: sources, hypotheses, findings

Sources, hypotheses and experiments on their own sheets, with the key concepts shared by all.

  1. Sheets Sources, Hypotheses, Experiments, Findings.
  2. A source is a node with the attributes Link and Date; the quote goes into the body.
  3. Hypotheses connect to sources with “supports” and “contradicts”, experiments with “depends on”.
  4. A key concept needed on several sheets gets + another sheet…, not a copy: the gold line shows all its appearances.
  5. Contents shows which sheets are weakly linked, and Stack shows how a concept passes through the layers.

The website has the example A small research project — a ready-made start.

See also: One entity on several sheets, Contents

Prepare for a meeting in 15 minutes

A meeting sheet gathers the questions, and the frame brings in their context from other sheets.

  1. A new sheet Meeting with a node for each question to settle.
  2. Connect each question to nodes on the topic sheets: the Meeting frame shows the related material at once.
  3. Fill in a Decision attribute (text) as the meeting goes; an unfolded card (↘) shows it right on the cards.
  4. A saved view Meeting: open — the filter Decision: empty.
  5. After the meeting, Project → Download .plyloom, or export .drawio for colleagues.

See also: Saved views and smart sheets, Export

A new version of a document: what we must do

A change stands on the changes sheet and next to the processes it affects; the memo, the checklist and the matrix come from the same map.

  1. Open the example A new version of the returns policy (Project → Examples), or start from the task A new version: what we must do in AI → A new map from materials.
  2. Check every change against the clause numbers in its attributes: a draft from an AI is checked against the document, not taken on trust.
  3. On the sheet What to do, fill in the owner, the due date and the status.
  4. Project → Assemble a document: the sheets of changes, processes, actions and questions make the impact memo; save the settings as the assembly Memo for the manager. Assemble a table → the nodes of What to do make a checklist for spreadsheets; the matrix Changes × Affected processes shows what touches what.
  5. When the next version comes out, compare, accept the edits item by item and assemble again: Project → Assemble again: Memo for the manager.

Keep your own assessments on a private sheet: they go neither to an AI nor into the memo.

See also: Assemble: a document, a table, a diagram, Private: what never leaves the map, Edits item by item: comparing and AI answers

A CV: one achievement, many applications

Achievements are written once and stand on the sheets of different CVs. Change a number and assemble every version again.

  1. Open the example A CV for each vacancy. Each achievement is a node with the attributes Result in numbers and When.
  2. For a new vacancy, a new sheet CV: company; drag your profile, jobs and the fitting achievements onto it. They are the same nodes, not copies.
  3. With an AI: the task Fit a vacancy — the AI only chooses from the map and marks with questions what has no evidence.
  4. Project → Assemble a document → Word, the current sheet; turn off The node type above its name and Relations, include private sheets and nodes — your contacts go into the CV, while the private attribute Evidence does not. Save it as the assembly CV: company — the next version is one button.

Contacts are a private node: they never go to an AI, and go into the CV when you tick them in.

See also: Assemble: a document, a table, a diagram, Private: what never leaves the map, One entity on several sheets

Back to sections