The Binse Diagrams user guide.

Copyright 2026 Binse Analytics. All rights reserved.

Binse Diagrams draws things: flowcharts, processes, org charts, mind maps, database and architecture diagrams, network and rack layouts, floor plans, wireframes, whiteboards and workshop boards. It is one of the workspaces inside Binse, beside Data Flow, Dashboards and Schedules, and it opens from the home page or straight at http://127.0.0.1:8767/diagrams/.

Two things make it different from a drawing program. Connectors stay attached: glue a line to a shape and it follows, routing itself around whatever is in the way. And a diagram can come from data - from a spreadsheet, from a database you already connected in Binse, or from a Data Flow workflow - so a picture of your systems can be built from the systems rather than drawn from memory.

Your diagrams are files on your own disk. A .binsediagram is a zip in Documents\Binse\Diagrams, or in a synced folder, or in a repository. Nothing is uploaded and nothing phones home.

Contents

• Getting started
• Shapes and libraries
• Connectors
• Text and styling
• Frames, pages and layers
• Templates
• Automatic layout
• Data
• Diagrams as text
• Importing and exporting
• Process modelling
• Precision drawing
• Whiteboard and workshops
• Presenting
• Working together (Binse Enterprise)
• The assistant (Binse Enterprise)
• Accessibility
• Automation
• Keyboard shortcuts
• Limits
• What Diagrams does not do
• Troubleshooting
• Glossary

Two editions, and how this guide marks them

Diagrams itself is in both editions. What Binse Enterprise adds is inside it: working together on one diagram at the same time, comments, version history and branches, share links, the administration console, the assistant, and the tool server for outside agents.

Two chapters are marked (Binse Enterprise) in their heading, and a few paragraphs elsewhere start with Binse Enterprise. in bold. Everything not marked that way is in both editions. In the Binse edition the Enterprise parts are not greyed out - they are simply not there - and entering a Binse Enterprise licence key turns them on without restarting the app.

Getting started

1. Open Binse and choose Diagrams, or go to http://127.0.0.1:8767/diagrams/.
2. You land on a blank page with the shape palette on the left, the tool bar across the top of the canvas, and the properties panel on the right.
3. Press R and drag on the canvas: that is a rectangle. Type while it is selected and the words go inside it.
4. Hover the shape. Four small arrows appear around it. Click one and you get a second shape, the same size, already connected to the first. Click another arrow on that one, and again: a flowchart is a dozen clicks.
5. Press Ctrl + S. The first save asks for a name; after that it saves in place.
6. Press Ctrl + 0 to fit what you have drawn on the screen.

The canvas

The page is infinite unless you give it a size (Properties > Page size). Move around with the wheel, a trackpad, a pinch, the arrow keys, or by holding Space and dragging. Ctrl and the wheel zooms at the pointer.

Drag with the left button on empty canvas and you get a selection rectangle; it takes what is completely inside it, or, holding Alt, everything it touches. Shift and a click adds to the selection. Escape deselects.

Every drag has the same three modifiers, everywhere in the app:

As you move something, smart guides appear: dashed lines when an edge or a centre lines up with something else, and little distance labels when the gaps either side are equal. They work at the same tolerance whatever the zoom, so they do not become useless when you zoom out.

Undo

Ctrl + Z undoes, Ctrl + Y (or Ctrl + Shift + Z) redoes. One gesture is one step: a drag that moves nine shapes undoes in one, and typing a word undoes as a word rather than a letter at a time. Applying a layout, a find-and-replace across every page, and an assistant's proposal are each one step too. Each open diagram has its own undo history, so undoing in one tab never reaches into another.

Saving, and what a diagram is

Your diagrams are files. A .binsediagram is an ordinary zip: a document.json with the diagram, the pictures you put in it, and a thumbnail. You can open it, copy it, put it in a synced folder or commit it to a repository. Nothing about it needs Binse to be running.

• Save (Ctrl + S) writes into Documents\Binse\Diagrams, or wherever you last saved.
• Save a copy to this computer (Ctrl + Shift + S) writes it anywhere, including a mounted network or cloud folder: if the folder appears in Explorer, it works.
• Open (Ctrl + O) lists what is in the Diagrams folder and offers to open a file from anywhere else. A file from outside the folder opens as a copy, so saving it asks where to put it. Whatever you open arrives in a tab of its own, beside what was already open.
• Double-clicking a .binsediagram or a .bdx in Explorer opens it in Binse: in a new tab of the Diagrams window if one is open already.
• Protect the diagram with a passphrase (Ctrl + Alt + Shift + P) encrypts the file. There is no way back into a protected file without the passphrase - not by us, not by anybody.

Autosave keeps the diagram's own file up to date. The switch beside the diagram's name turns it on: the file is rewritten 2 seconds after you stop making changes, and a change never waits longer than 30 seconds however fast you are working. The switch says when it last wrote. It is your own setting rather than the diagram's, so it stays as you leave it for the next diagram you open.

Two things it deliberately does not do. It never saves a diagram it would have to ask you about first, so a diagram that has never been saved, or one that is only a file on this computer, leaves the switch off and amber with the reason on it; click it, save the diagram into Binse once, and autosave starts. And a save that fails switches autosave off and tells you why rather than trying again quietly. Your changes are still there and still unsaved, and the close guard still stops you walking away from them. Nothing is more dangerous than a switch that says your work is being saved while the folder it writes to has gone.

Recovery copies are the other half, and they stay. While there is unsaved work, a copy of the diagram is written every 30 seconds to a folder of its own. If the machine or the browser stops, the next start offers the copy back and says how old it is. A clean save deletes the copy, so with autosave on there is rarely one to offer: every couple of seconds your work is already in your own file, which is the better place for it.

Several diagrams at once

Every diagram you open, start or import gets a tab of its own in the strip over the canvas, the way a browser or an office program keeps documents side by side. Opening a diagram never replaces the one you were working on, so nothing asks whether to throw your changes away. The one exception is the blank diagram the app starts with: if you have not touched it, the first diagram you open takes its place rather than leaving an empty tab behind.

• Click a tab to bring its diagram forward, or use Ctrl + Alt + Page Down and Ctrl + Alt + Page Up. The arrow keys move along the strip when a tab has the focus, and the command palette lists every open diagram by name.
• Drag a tab to put it somewhere else, or right-click it for Move left and Move right. Alt + Shift with an arrow moves the tab that has the focus.
• Close a tab with its cross, a middle click, or Ctrl + Alt + W. A diagram with unsaved changes asks first: Save, Don't save or Cancel. Right-click a tab and choose Close the others to close everything else, one question at a time for anything unsaved. Closing the last tab leaves a blank diagram.
• The + at the end of the strip starts a new blank diagram. A dot on a tab means that diagram has changes that are not in its file yet.

Each tab keeps its own state. Its undo history, what is selected, which page is showing, where you had scrolled and zoomed to, its passphrase, and its autosave and recovery copies all stay with it while another tab is in front. Autosave writes each diagram to its own file; a change still waiting to be written when you switch away is written as you switch.

Copy and paste works across tabs. Shapes arrive with their named styles, their pictures and their links to data: a shape bound to a row of a spreadsheet or a query is still bound in the other diagram, and the data source comes along with it. One undo takes the paste away, data source and all.

Opening the same diagram twice brings its tab forward instead of opening a second copy. And a diagram sent from Data Flow (Send to Diagrams) opens in a tab of its own too, in the Diagrams window that is already open if there is one.

In Binse Enterprise, a shared diagram's live session follows the tab in front: bringing another tab forward leaves the shared session, and coming back joins it again.

Moving between workspaces

The binse Diagrams wordmark at the top left goes back to the home page, and the workspace name beside it opens the list of the other workspaces.

Shapes and libraries

The palette on the left holds 10,374 shape definitions in 190 libraries, and beside them 111,893 icons in 1,045 packs. The search box at the top searches all 122,267 of them at once, by name, by the words in the definition, and by tag.

• Click a shape in the palette and then click on the canvas to place it, or drag it out.
• Drop a shape onto a connector and it is spliced into the line.
• Star a shape in its preview to put it in Favourites, at the top of the palette.
• Recents remembers what you last used, per diagram.
• The libraries are grouped by what they are for, and a group collapses with a click.

What is in the box

Behind the 34 icon sets in the palette sit 1,045 packs of 111,893 icons, which the search box reaches whether or not their pack is open.

An icon is a shape in its own right, and it is also something a shape can wear: any shape has an Icon in the properties panel, with a placement, a size and a colour.

Shapes of your own

• Make a shape from a selection or a picture. Select anything and run Create a shape from the selection or a picture file from the command palette (Ctrl + K). It joins My libraries and behaves like any other definition. A .svg becomes editable paths; a .png becomes a shape with the picture inside it.
• Import shapes from a stencil file (.vssx) into a library of your own. The report afterwards says what came across and what was approximated.
• A plugin can add a library. See Automation.

Your own libraries live in Documents\Binse\Diagrams\Libraries as JSON files, so they can be copied to a colleague or kept in a repository.

Connectors

Press L, or hover a shape and drag from one of the blue dots on its edge. Let go on another shape and the line is glued: move either shape and the line follows, re-routing as it goes.

There are two kinds of glue, and the difference matters:

• Glued to the shape. The line picks whichever side is sensible and changes its mind as things move. This is the default, and it is what you want almost always.
• Glued to a point. Drop on one of the small marks on the outline and the line stays on that exact point.

Routing

A connector is orthogonal by default: right angles, going around shapes rather than through them, re-routed while you drag rather than after you let go. The other shapes are:

Select a connector and its waypoints appear as handles, with a smaller handle at the middle of each segment: drag a waypoint to move it, drag a middle to add one. A segment you move is pinned and the router leaves it alone afterwards. Reset route (in the command palette) clears every waypoint and pin and lets the router start again. Where two lines cross, the one on top hops over - Jump over crossings turns that off.

Ends, styles and labels

There are 20 arrowheads, available at either end: none, arrow, open, stealth, triangle, diamond, diamond open, circle, circle open, bar, cross, crowfoot many, crowfoot one, crowfoot one many, crowfoot zero one, crowfoot zero many, uml inherit, uml aggregate, uml compose, uml dependency.

The line itself has a width, a dash pattern, a cap and a join, and can be animated so the flow runs along it (on screen, in the web page export, and in an animated GIF - not in a still picture).

Any number of labels can sit on one connector. Double-click the line to add one, then drag it along the line; it keeps its position along the line as the line changes. A label has its own text styling and its own background.

For a line that leaves the page, use an off-page connector: two shapes that name each other, one on each page, each a link to the other.

Building from the keyboard

With a shape selected, Tab adds the next shape at the same level and Enter adds a child, both connected - which is how a tree or a mind map is fastest to build. Alt + Shift + an arrow adds a connected shape in that direction. Alt + C starts a connection and then arrow keys or Tab pick what to connect to, Enter confirms.

Text and styling

Double-click anything to type in it, or press Enter with it selected. Text inside a shape is rich: bold, italic, underline, strikethrough, code, colour, size, font, highlight and links, with alignment, bullet and numbered lists and indenting.

A text box can wrap, shrink to fit or grow as you type; a shape can grow with its text. Text can also run along a path, and can be converted to outlines when you need the letters to be shapes.

Find and replace (Ctrl + H) works across every page and every label, and the whole replacement is one undo.

Spell check (F7) underlines what it does not know and keeps a dictionary of your own words.

International text

Right-to-left text (Arabic, Hebrew) lays out right to left, including inside shapes and on lines, and mixes with left-to-right text in one paragraph. Chinese, Japanese and Korean text breaks between characters rather than at spaces, and can be set vertically, top to bottom in columns. Set the direction per text block in the properties panel; a diagram can hold several.

Maths is written between dollar signs - $E = mc^2$ - and drawn as maths, in a shape's text or in a label.

Fills, strokes and effects

A shape's fill is a solid colour, a linear or radial gradient, a pattern, or nothing; its stroke has a colour, a width, a dash pattern and an opacity. Beyond that there are shadows, blend modes, blur, pixelate and magnify effects, and, for a picture, masking and adjustment.

Named styles

Give a set of properties a name - New style from this in the properties panel - and apply it to anything. Change the named style and everything wearing it changes. Detach takes a shape off the style, keeping how it looks.

The format painter (Ctrl + Shift + C) copies the look of one shape onto the next thing you click. Hold Shift as you click and it keeps painting until you press Escape.

Select similar items (Ctrl + Shift + A) selects everything that looks like, or is the same kind of thing as, what you have selected - which is how you restyle every decision on a page at once.

The eyedropper (Alt + I) takes a colour from anywhere on the canvas.

Themes, light and dark

A diagram has a theme: a small set of named colours ($accent-1, $shape-fill, $border and so on) with a light and a dark variant. Shapes refer to the names, so the same diagram looks right in both. The theme editor changes the variants, and an export can be taken in either.

The app has its own light and dark setting (Alt + Shift + D), independent of the diagram's.

Hand-drawn look (in Properties > Hand-drawn look) redraws the whole diagram as if sketched. The wobble comes from a fixed seed kept with the diagram, so an export looks exactly like the screen and looks the same tomorrow.

Fonts

The fonts that ship with the app travel inside exports, so a PDF or an SVG looks the same on a machine that does not have them. A font installed on your computer can be used as well - it will be embedded in an SVG or a PDF where its licence allows, and substituted where it does not, which the export report tells you about.

Frames, pages and layers

Frames and containers

A frame (F) is a labelled box that owns what is inside it. Drop something in and it becomes a child: move the frame and the contents move with it. A frame either clips what hangs over its edge or grows to fit. Frames nest.

A container is the same idea with a purpose: a pool, a lane, a group box, a rack. Dropping a shape into a lane reparents it in one operation, so it is in that lane for layout, for export and for the reading order.

Swimlanes are containers in a row or a column, each with a title. Resize one and the others follow. The swimlane layout keeps every shape in its own lane.

A layout frame arranges its own children: drop something in and everything re-arranges. See Automatic layout.

Pages

The tabs along the bottom are pages. Add, rename, reorder, duplicate and delete them there. Each page has its own size, background and grid. Ctrl + PageDown and Ctrl + PageUp move between them.

Layers

Each page has layers, listed in the panel. A layer is visible or hidden, locked or not, and printable or not - so a layer of notes to yourself can be on the screen and out of the export. Move a selection to a layer from its right-click menu.

Linked copies

Ctrl + Alt + D makes a linked duplicate: a copy that stays the same as its original. Edit the text, the style or the size of any of them and every copy changes. It is how a legend, a logo or a repeated sub-process stays in step across twenty pages. An ordinary duplicate (Ctrl + D) is not linked.

Behind the pages there is also an element model: one thing - a system, a service, a person - can appear on several pages as several shapes. Rename the element and every appearance is renamed. The Elements panel lists them and says where each one appears.

Templates

File > New diagram… (Alt + N) opens the gallery: 220 templates, with a search box and categories down the side. Click one to preview it, then Use this template.

Templates of your own

File > Save as template… puts the open diagram in My templates, with the name, description and category you give it. Your templates live in Documents\Binse\Diagrams\Templates and appear in the gallery beside the shipped ones. Delete one from its card.

A template is an ordinary .binsediagram, so a team template can simply be a file on a shared folder that everybody copies into their own Templates folder.

Automatic layout

Press Ctrl + Shift + L. The menu offers the whole page or the selection, suggests the algorithm that suits the shape of what you have drawn, and shows the settings for the one you pick. Apply layout is one step, and one undo puts everything back exactly where it was.

Common settings: the direction (down, right, up, left) for the ones that have one, the space between shapes and between ranks, whether the lines are re-routed afterwards, and stay close to the current positions, which keeps the drawing recognisable instead of rebuilding it.

Ctrl + Alt + L applies the suggested layout without opening the menu.

Layout frames

Turn a frame into a layout frame and it arranges its own children with the algorithm you choose, again whenever something is added, moved in or removed. It is how a diagram that grows - a list of services, a set of teams - stays tidy without anybody tidying it.

Mind-map mode

A mind map has its own keys, and they are the whole point:

• Enter adds a child of the selected topic.
• Tab adds a sibling after it.
• Alt + Shift + C collapses or expands a subtree, with a count on the collapsed one.

The branches re-balance themselves either side of the central idea as you type, so you can write a map without ever touching the mouse.

Data

A diagram can be bound to data: shapes take their text, colour, size, badges and icons from rows of a table, and change when the table changes. The Data tab of the panel is where all of it lives.

Data sources

Add a data source takes a spreadsheet (.xlsx), a .csv, a .tsv, a .json or a .parquet file; Tables from SQL runs a bounded query on any connection you have set up in Binse (the same connections Data Flow and Dashboards use); From a web service reads a web address as a table.

Each source has a key column - the column whose value identifies a row - and a refresh setting: manual (only when you ask), open (every time the diagram opens), schedule (on a schedule).

A web service source has a method, parameters, headers, an authentication method (none, api-key, basic, bearer, oauth2) and a way of paging through results (none, page, offset, cursor, link, next-url). Passwords and tokens are stored by the engine as secrets and are never written into the diagram, so a diagram with a web service in it can be sent to somebody safely.

Linking a shape to a row

Select a shape, open Data > Linked row, and pick the value of the key column. From then on the shape carries that row: {col} in its text is replaced by the column, the Shape data section shows every field, and rules can look at any of them.

Link many at once by matching a column against the shapes' text.

Shape data

Any shape can carry fields of its own - an owner, a cost, a status, a date - whether or not it is linked to a row. Fields are typed, they are searchable, formulas can read them, and they travel in the exports that have somewhere to put them.

Rules and data graphics

A rule watches a condition and changes something. Rules apply in order, and for one property the last matching rule wins.

A condition compares a field with a value using =, !=, <, <=, >, >=, contains, starts, ends, empty, not-empty, in, and several conditions combine with and or or.

The richer effects are data graphics:

A legend is generated from the rule set - every rule that has a label becomes an entry with its swatch and the number of shapes it currently matches - so a diagram coloured by rules explains itself.

A smart container groups the shapes it holds by a field, making a lane per value and moving shapes between lanes as the data changes.

Formulas

The Binse formula language - the same one Data Flow and Dashboards use - can be typed into any data field, and into text between {= and }. {=IF([Status] = "Late", "!", "")} in a shape's text puts an exclamation mark on the late ones. Variables (Data > Variables) are values the whole diagram can use, written {=var.name}.

Charts

13 kinds of chart are drawn natively - as shapes, not as pasted pictures, so they scale, theme and export like everything else: bar, column, line, area, pie, donut, scatter, gauge, kpi, sparkline, progress, table, radar.

A chart takes its categories and series from a data source, or from numbers you type. It can group, filter, aggregate, sort and limit before it draws.

Making a diagram from a table

Data > Make a diagram from data builds the diagram rather than decorating it:

• From a table - a flowchart or a cross-functional flowchart from a spreadsheet of steps.
• Org chart from a table - a person per row, with a column naming their manager.
• Tables from SQL - see below.
• A Data Flow workflow - see below.

Everything it makes stays linked, so Refresh brings changes across. And the link works both ways: write-back sends your edits on the canvas - a renamed step, a new box, a changed owner - back to the spreadsheet or the table, after showing you exactly what it will write.

Database diagrams from a Binse connection

Data > Tables from SQL… reads the schema of a connection you have already set up in Binse and draws it: a box per table with its columns, its primary and foreign keys marked, and a line per relationship in the notation you choose (crowfoot, idef1x, chen).

Pick the tables you want rather than all of them; a hundred-table database is not a diagram anybody reads.

Refresh goes back to the database and shows a schema diff: tables and columns added, removed and changed since you drew it, each with a tick, so you choose what to bring across.

Going the other way, Export the tables as SQL writes CREATE TABLE statements in 9 dialects: SQLite, PostgreSQL, MySQL / MariaDB, SQL Server, Snowflake, BigQuery, Databricks, Oracle, Standard SQL.

A Data Flow workflow as lineage

Data > A Data Flow workflow opens one of your Binse workflows as a lineage diagram: a shape per tool, lines for the data between them, and columns traced through the joins, selects and renames, so you can see where a field came from and what happens to it downstream. It reads the same Documents\Binse\Workflows folder the Data Flow canvas does.

The lineage diagram is a diagram like any other - lay it out, style it, put it in a document - and refreshing it re-reads the workflow.

Diagrams as text

Every diagram has a text form, .bdx, which is lossless: save as text, open it again, and you have exactly the diagram back. Ctrl + Alt + T splits the window - text on one side, canvas on the other - and edits on either side appear on the other.

@diagram "Order processing"
@page "Page 1"
  start: "" { shape: uml.initial-node }
  receive: "Order received" { shape: uml.action }
  check: "Valid?" { shape: uml.decision }
  line = start -> receive
  line_2 = receive -> check
  reject_line = check -> reject: "[invalid]"

The text form takes @import of other .bdx files, variables, and glob selectors that style many shapes at once - so a large drawing can be several files that compose into one.

Mermaid, PlantUML and D2

Diagram-description languages are read and written as well:

• Mermaid text (.mmd), 20 diagram types.
• PlantUML text (.puml), every family.
• D2 text (.d2), the full language.

Paste any of them onto the canvas and it is drawn. Export back to any of them, with a report of anything that had no equivalent.

Live code frames

A code frame keeps its source with it. Put Mermaid, PlantUML or D2 text in a code frame and the frame redraws whenever the text changes. Edits you made by hand to the drawn shapes are kept where they still make sense, and the drawing you replaced is offered back in one click if the new one is not what you wanted.

Importing and exporting

Importing

Drop a file on the canvas, use File > Import a file… (Ctrl + Shift + O), or paste diagram text. 39 formats are read:

Every import ends with a report saying what came across, what was approximated and what was left out, shape by shape. Read it: a drawing from another program is never a perfect round trip, and the report is where the guide stops guessing and the file tells you.

Importers treat their input as hostile. Nothing is fetched from the network, external entities are refused, and the size, depth and expansion limits in Limits are checked before anything is expanded, so a malicious file is refused rather than survived.

A network from IT's own files

File > Network from device files… draws a network diagram from files IT already has, several at once: a device list (CSV or Excel, with whatever column names the list uses - name, type, address, MAC, model, maker, site, rack, position, height), network scan results (the XML a port scanner writes), the LLDP and CDP neighbour tables a switch, router or server prints (the one-line table and the detail listing, pasted or saved to a text file), DHCP lease exports (a Windows server's lease export, the Unix lease file, the one-line lease file) and ARP tables (from a router, from arp -a on Windows, Linux or a Mac, from ip neigh). Diagrams reads files only: it never scans the network, because a scan from a managed laptop can set off a company's security alarms.

Each file is recognised from its contents; change Read as when the guess is wrong. A neighbour or ARP table is named after the device it was printed on - from the prompt in the file, else from the file name - and Taken on lets you say so when neither does.

The same device in several files becomes one shape (matched by MAC, then name, then address), shown with the network library's symbol for its kind - firewall, router, switch, access point, server, computer, phone, printer, camera - and its address, MAC, model, open ports and ports in use in its shape data. Links come from the neighbour tables, labelled with the port at each end. By layer puts firewalls and routers at the top, then switches, then servers and access points; by site draws a frame per site. Computers and phones seen only in ARP and lease tables are grouped in a frame per subnet, under the device that routes it. When the device list gives racks and positions, a Racks page draws each cabinet with its devices at their units.

Updating. Run the same command on a diagram that already has a network and it offers to update it from these files instead. Devices whose address, MAC, model, operating system, open ports, site or rack changed get a Changed badge and the change in their "Changes" field; new ones are added with a New badge, next to the device they connect to; ones that none of the files mention get a Missing badge and stay where they are. Nothing is deleted, and your own fields, styles, moves and edited labels are kept. The update is one step: Undo takes it all back.

Exporting

File > Export… (Ctrl + Shift + E). Choose the format on the left and its settings on the right; the preview underneath is what you will get.

32 formats are written:

Common settings: this page or every page, the light or dark theme, the scale, whether the background is included, and the paper size and orientation for the ones on paper.

Export every frame or page as a file (Ctrl + Alt + Shift + E) writes one file per frame or per page in one go, named after each.

The web page export

Web page writes one self-contained .html file: no server, no internet, nothing to install. It pans and zooms, follows the links in the diagram, honours show-and-hide actions and hotspots, and carries every page of a multi-page diagram. It is the easiest way to give a diagram to somebody who has no Binse.

The exported page has a few keys of its own: + and - to zoom, 0 to fit, 1 for actual size.

Web page with a picture map writes a plain picture with an HTML image map over it, for putting in a page that will not run scripts.

Printing

Ctrl + P. Choose the sheet size and orientation, fit to a number of pages or print at a scale, set an overlap for tiling a big drawing across several sheets, and add headers and footers. The preview can be panned, and it shows the page breaks on the canvas while the dialog is open.

Process modelling

BPMN

The BPMN 2.0 library holds every element of the notation, and the check (Ctrl + Alt + B) reads the model structurally: a start that leads nowhere, an activity with no way out, a gateway that never closes, a message flow inside one pool. Click a problem and it selects the shape.

A sub-process drills down: double-click it and you are on its own page, with a way back up.

Modelling conventions

Beyond what BPMN requires, a team usually has rules of its own. The Process tab of the panel has a Conventions section that turns them on and shows hints on the canvas as you draw.

Three presets set a sensible group of these at once: Descriptive (6 rules), Analytic (12 rules), Strict (17 rules).

Simulation

Token simulation plays the process: tokens move along the sequence flows, you can play, pause and step, and you watch where they pile up.

Process simulation is the quantitative one. Give the activities durations, the gateways probabilities, the roles a number of people and a working calendar, and a cost per hour, then run it: the result is time per case, queue lengths, resource use and cost, with charts, and what-if scenarios side by side.

Decisions (DMN)

A decision table has inputs, outputs, rules as rows, and a hit policy. The decision requirements diagram shows which decisions feed which. A table is validated as you edit it - overlapping rules, gaps, unreachable rows - and exports as .dmn.

Publishing

Process documentation turns the model into a document: every activity with its description, owner, inputs and outputs, every decision, every role, with the diagram at the front. It publishes as a document (.docx), a spreadsheet (.xlsx) or a small static web site.

Precision drawing

Units and scale

Properties > Scale and units sets the unit (millimetres, centimetres, metres, inches, feet, or plain page units) and the drawing scale (1:50, 1:100, ¼" = 1'). Every length, every dimension line and every room area is then given in real units, and changing the scale re-reads them all.

Rulers (Alt + R) run along the top and the left, with a movable origin, and you drag guides off them.

The pen and paths

The pen (P) draws a path: click for a corner, drag for a curve, Enter finishes, Escape cancels. Ctrl + Alt + P opens a finished path for point editing - move points, change the handles, add and remove points, and switch a point between corner and smooth.

Paths can be joined, split, reversed, and a stroke can be outlined into a filled shape.

Booleans

Unite (Ctrl + Alt + U), subtract (Ctrl + Alt + S), intersect (Ctrl + Alt + I) and exclude (Ctrl + Alt + X) combine shapes. The result stays editable: the shapes that went in are still there inside it, and moving one re-computes the result. Expand (Ctrl + Alt + E) flattens it into an ordinary path when you want to stop.

Walls, rooms and doors

The wall tool (W) draws walls that mitre where they meet, with a thickness you set. Type a length as you draw to make it exact. Doors and windows are hosted by a wall: they cut their own opening and move with it. Close a run of walls and you get a room, with its area worked out and kept up to date.

Dimensions and snapping

The dimension tool (N) measures between two points and stays attached: move what it measures and the number changes. Linear, aligned, angular and radial dimensions, with the precision and units you set.

Object snaps (F3) catch ends, midpoints, centres, intersections and perpendiculars while you draw; Tab cycles through the candidates when several are close. There is also the ordinary grid snap and the smart guides.

CAD drawings go both ways: .dxf in and .dxf out, with layers, blocks and real units.

Quantities and costs

The Costs tab of the side panel turns a floor or site plan into a costed list of materials. Each diagram has its own price list: items with a name, a unit (each, m, m² or m³), a price and a currency, and optionally a waste percentage, a depth or height, and a code of your own.

An item is matched to what it prices. Select something on the plan - a desk, a wall, a room - and Add items for the selection makes one item for each kind of thing selected, already matched: every desk on the page is then the desk item, now and later. An item can also match every shape that carries a tag, so one kind of shape can be priced two ways.

The quantity is measured from the drawing at the page's scale: furniture is counted, walls, fences and kerbs are measured along their length, rooms, floors and paving over their area, and a volume is an area times the item's depth. Waste is added before the price is applied; counted items round up to whole ones. A page with no drawing scale is flagged, with a button to set one.

Put a schedule on the page draws the quantities and costs as a table beside the plan, with a total per currency. It follows every change to the drawing and the prices, and Undo takes an edit and its costing back together.

Read a price list… takes a CSV or Excel file, finding the columns by their headings (Item, Unit, Price, Currency and the like, in any of the interface's languages) and saying which rows it left out. A list can be saved under a name and used on other drawings, and the list and the page's quantities can be saved as CSV or Excel. Use the costs in Dashboards adds the saved diagram to Dashboards as a dataset of its costed lines, which follows every save.

Whiteboard and workshops

The whiteboard tools are on the same canvas as everything else - a workshop board can become a process diagram without moving anything to a different app.

Sticky notes and ink

Sticky notes (S) come in six colours (Alt + 1 to Alt + 6), carry an author, tags and votes, and grow their text to fit. Tab while writing one adds the next.

Pen (B), highlighter (I), marker (M) and eraser (E) draw ink, with pressure and tilt from a pen. Turn the selected ink into shapes (Shift + X) recognises rectangles, ellipses, diamonds, arrows and lines and replaces the ink with real shapes, glued where they touch.

The lasso (Q) selects around an awkward group. The freehand connector (A) draws a line between two shapes and glues it. Fill (K) pours colour into an enclosed area. Stamps (J) are stickers, washi tape, marks and animated pictures.

Tables, code and notes

A table (U) has rows and columns you can navigate with the arrow keys, Tab and Enter, and takes CSV in and out. A code block shows highlighted source. A note takes Markdown. An embed shows an offline poster and loads nothing at all until somebody asks it to.

Workshop objects

Alt + W opens the workshop panel: voting with a number of votes each, polls, estimation and card sorting, plus a timer everybody can see. Meeting and agenda frames, kanban and timeline frames give a session its shape.

Cluster and sort groups the notes on a board by colour, tag, author or keyword in one step.

Binse Enterprise. In the Binse edition, votes, polls and the timer work for one person on one machine - which is enough for a board you are preparing, or a session where one person drives the screen. Running them live with other people, each on their own machine, is part of working together.

Tablet and pen

Touch and pen layout (from the command palette) switches to bigger targets with the tools where a hand can reach them, and palm rejection while a pen is in use.

Presenting

Frames or pages become slides. Alt + P opens the slide builder: add every frame, or the selected one, set the order by dragging, give each a title and notes, choose a transition, and set an autoplay time.

F5 presents from the first slide, Shift + F5 from the current one. While presenting there is a laser and a spotlight, and the arrow keys move between slides.

Presenter view is a second window: the current slide, the next one, your notes and a clock.

Interactive actions

A shape can do something when it is clicked - while presenting, in the exported web page, and in embed view mode:

• Go to a page, or to another diagram.
• Show or hide a layer.
• Zoom to a shape or a frame.
• Open an address.
• Run one of your scripts.

Hotspots are invisible regions with an action, for making part of a picture clickable. Collapse and expand (Alt + Shift + C) folds a subtree away, which is how a large tree is presented a level at a time.

Working together (Binse Enterprise)

Binse Enterprise. Everything in this chapter needs a Binse Enterprise licence and a relay - one server your organisation runs. Your administrator sets its address once; after that the app signs you in and the rest is automatic. Without a relay, Diagrams works exactly as the rest of this guide describes, on your own files.

Live editing

Open a shared diagram and you see who else is in it, where their pointer is and what they have selected. Edits arrive as they are made. If your connection drops, you keep working; your changes queue and merge when it comes back, and the merge is by operation rather than by file, so two people editing different shapes never overwrite each other.

Follow somebody and your view goes where theirs goes.

Comments

Press C for the comment tool and click a shape or a point. A comment is a thread: replies, mentions, reactions, and Resolve when it is dealt with. The comments panel lists them, filtered by page and by whether they are resolved. / opens the session chat.

Comments are written into the diagram file, so they travel with it and are still there for somebody who opens it without a relay.

Version history and branches

Every save is a version. The history panel lists them with who saved and when, lets you name the ones worth naming, compare any two - added, changed and removed, shown on the canvas - and restore one, as a new version rather than by destroying what came after.

A branch is a copy that knows where it came from: work on a change without disturbing the diagram everybody is using, then merge it back through a review that lists every difference.

Sharing

A share link carries a role - view, comment or edit - and can have an expiry date, a passphrase and be revoked at any time. A guest is somebody outside the organisation who has a link and no account.

Administration and audit

The administration console sets the organisation's policy: who may share outside, whether guests are allowed, how long links live, which workspaces exist. Everything that matters is written to an append-only audit log: who opened, shared, exported, restored, merged and revoked.

A diagram can be end-to-end encrypted, in which case the relay stores and forwards ciphertext and can read nothing at all. Version comparison, comments and live editing still work - they happen in the browsers - but server-side search does not, because there is nothing to search.

The assistant (Binse Enterprise)

Binse Enterprise. The assistant is in Binse Enterprise only. In the Binse edition it is not offered, not gated: the tab, the commands and the shortcut do not exist.

Ctrl + I opens the Assistant tab, with whatever you have selected as its context.

What it can do

• Generate a diagram from a description - flowchart, sequence, org chart, entity relationship, mind map, BPMN process, architecture.
• Edit by instruction - "make every decision orange", "add error handling paths".
• Explain or summarise the diagram in words.
• Write alternative text for the page and for anything that has none.
• Suggest a layout.
• Cluster and label sticky notes.
• Build a diagram from a pasted document or table, from a PDF, a word-processor document, a spreadsheet or a slide deck, or from a photograph of a whiteboard - sticky notes come back as sticky notes, keeping their colours and their grouping.
• Write rules and formatting from a description - "colour red when the target is missed" - which you can then edit in the Data panel like any other rule.
• Fix validation problems it finds.
• Answer "how do I…?" from this guide and the other documentation shipped with your copy, and offer to run the command for you.
• Save a template or a slide deck, or make a wireframe into a working web page.

It never generates pictures.

Proposals, and why nothing changes behind your back

The assistant never edits your diagram directly. Everything it does arrives as a proposal:

• On the canvas, additions are ghosted in green, changes outlined in amber, removals struck through in red.
• In the panel, every change is listed with its own tick box.
• Apply writes the ticked changes as one transaction - one Ctrl + Z takes all of it back.
• Reject leaves the diagram exactly as it was.

Changes that depend on each other tick and untick together, so you cannot apply a connector to a shape you did not accept.

Shapes appear in the proposal while the answer is still being written, so you can see where a large edit is going.

Your key, or the organisation's

Assistant > Settings has two ways of paying for it:

• Your own key. Paste it into the field labelled Anthropic API key (for Claude models). It goes to the engine on this computer, is stored by Windows for your account only, and is never sent back to the browser or written to a log.
• The organisation's. Your administrator gives you an address and a token; usage is counted and capped centrally.

There is a monthly spend cap (a default of $20, which you can change), and the settings show this month's usage and the recent requests with what each cost.

What is sent

With each instruction: the outline of the current page - the text, kinds and connections of its shapes - the page names, what you have selected, and what you typed or pasted. During a run the assistant can read more of this diagram.

Never sent: other diagrams, files on your computer, or anything outside the document. On a large diagram the outline is summarised and the panel tells you that it was. The settings panel shows the exact text that will be sent.

Accessibility

Everything from the keyboard

Every action has a keyboard route, including drawing a shape and connecting it, which is unusual for a drawing program and is deliberate here.

• F6 and Shift + F6 move between the areas of the window: tools, shapes, canvas, panel, pages.
• Tab and the arrow keys move around the canvas; Alt + an arrow selects the nearest shape in that direction.
• Ctrl + Enter inserts the current tool's shape in the middle of the view.
• Alt + Shift + an arrow adds a connected shape.
• Ctrl + an arrow resizes; Ctrl + Alt + an arrow rotates.
• F2 edits the text, title or label of whatever is selected.

The outline

Alt + O opens the outline: the diagram as a list, in reading order, which is what a screen reader announces. It says what each shape is, what it says and what it connects to. You can set the reading order here with Alt and the arrow keys, and pressing Enter shows the shape on the canvas.

The checker

Alt + A checks the page, or every page, and lists what it found with a fix for each:

Contrast is measured against the actual background, in both the light and the dark variant of the theme, and text counts as large - and so is allowed 3:1 rather than 4.5:1 - at 24 units, or 18.66 units bold. Text under 10 units is reported whatever its contrast.

Every shape definition that ships with the app passes in both themes, and every shipped template reports no errors in either.

High contrast and motion

Alt + Shift + H turns on a high-contrast rendering. If your system asks for reduced motion, the app does not animate: transitions, the layout animation and the presentation transitions all become instant.

Automation

Plugins

A plugin can add shape libraries, tools, importers, exporters, data sources, panels, commands and shapes drawn by code.

Install from the Plugins panel: a folder or a zip. A plugin arrives switched off. Turning it on shows exactly what it is asking for - to read the diagram, to change it, to add libraries, to add formats, to add interface - and it gets nothing you do not grant. Plugins live in Documents\Binse\Diagrams\Plugins.

Scripts

The Scripts panel runs JavaScript against the open diagram: query it, change it, measure it. Ctrl + Enter runs, and a script can be saved into the diagram or attached to a shape so that clicking the shape runs it while presenting.

const ids = await api.query({ type: "shape", text: "TODO" });
await api.setStyle(ids, { fill: "#fde68a" });
return ids.length;

A script can read and change the open diagram and nothing else: no network, no files, no registrations. A script that runs for more than 30 seconds is stopped.

The command line

binsediagrams renders, converts, validates, lints and watches, from a shell or a build:

binsediagrams render <file|folder>... [--out <dir|file>] [--format svg|png|pdf|…] [--page <n|name|all>]
                                      [--theme light|dark] [--scale <n>] [--no-background] [--quiet]
binsediagrams convert <input> <output> [--format <id>] [--page <n|name>] [--theme light|dark]
binsediagrams validate <file|folder>... [--json]
binsediagrams lint <file|folder>... [--json] [--max-warnings <n>] [--no-accessibility]
binsediagrams watch <file|folder>... [--out <dir>] [--format svg|png|…]
binsediagrams formats

It exits 0 when all is well, 1 when validate or lint found problems, and 2 on a usage or file error - which is what makes it useful in a build that should fail on a broken diagram. SVG from the command line is byte-for-byte the same as SVG from the app.

The local API

The engine offers a small REST API on this computer for scripts and tools: list, read, write, export, validate and search your diagrams, and read and write their comments.

Three locks, all on at once:

1. A key. Make one in Plugins > Local API > Create key. It is shown once. Send it as Authorization: Bearer bdk_…. A key is read-only or read-and-write, and you can revoke it.
2. This computer only. The engine listens on the loopback address and refuses any other caller.
3. Not from web pages. Any request that comes from a page is refused outright.

The full description of every route is served by the engine itself at /api/v1/openapi.json. Webhooks call an address of yours when a diagram is saved.

Embedding

A built copy can be embedded in a page of your own in view or edit mode and driven through messages - load a diagram, export one, listen for changes.

The tool server

Binse Enterprise. A tool server lets an outside agent work on your diagrams through the same operations the app uses: read the outline, query shapes, add and change them, lay a page out, import text. It is a separate program, started deliberately, and it is in the Binse Enterprise installer only.

Keyboard shortcuts

Every one of the 139 shortcuts the app answers to. Ctrl is Cmd on a Mac keyboard.

Tools

Whiteboard

Precision

Edit

Select and move

Keyboard

Arrange

Text

Insert

Pages

View

Present

File

While the pointer is down

In a list, a menu or a table

Panels, menus and whiteboard tables all answer to the same keys: the arrow keys move, Home and End jump to the ends, Enter or Space chooses, Escape closes, and Tab moves on. They are not listed per panel because they are the same everywhere.

If these are not the letters you want

Settings (Ctrl + ,) has a Keyboard section: alternative letter layouts, and your own bindings for anything.

Only the bare tool letters move between layouts; everything with a modifier is the same in all of them. Your own bindings are applied on top, and the settings warn you when two commands answer to the same keys.

The command palette

Ctrl + K searches every command in the app by name, by group and by what it does, and shows the keys for the ones that have them. It is the fastest way to reach anything you do not use often enough to remember.

Limits

They are limits on what is read and made rather than on what you may do: they exist so that a malformed or hostile file is refused quickly instead of filling memory.

What Diagrams does not do

Honestly, and so you do not go looking:

• Binary .vsd (the old drawing format) is not read. .vsdx, .vdx, .vssx and .vstx are.
• .dwg is not read or written. .dxf is, both ways, and every CAD program writes .dxf.
• No picture generation. The assistant edits diagrams; it does not draw illustrations.
• No diagram visual inside Dashboards yet: a diagram exports as a picture or a web page, and a dashboard can show that, but a live diagram is not one of the Dashboards visuals.
• No viewer without a licence. Opening a .binsediagram needs a licensed copy. To give a diagram to somebody who has none, export a web page or a PDF.
• One diagram at a time in a window. Open a second window for a second diagram.
• The relay is not installed with the app. Working together needs a server your organisation runs; it ships as a separate zip with its own install guide.
• Cloud vendor icon packs are not shipped. The cloud library is vendor-neutral. You can import a vendor's own stencil file into a library of your own.
• No macOS or Linux application. Windows only for now.
• Shape, library and template names are in English in every language, and so is search over them; so are exported files and the pictures in this guide. The rest of the app can be in English, French, German, Spanish, Portuguese, Italian or Dutch: Settings > Appearance > Language.

Troubleshooting

A diagram will not open, or opens as a copy. A file outside Documents\Binse\Diagrams opens as a copy with nowhere to save back to; use Save a copy to this computer. A protected file needs its passphrase, and there is no way around it.

An import lost something. Read the import report - it lists what was approximated and what was dropped, shape by shape. A drawing that leans on one program's own features (custom code inside shapes, that program's own data links) will not come across whole; the geometry, the text, the glue and the layout do.

An export has the wrong fonts. A font whose licence forbids embedding is substituted, and the export report says which. Use one of the fonts that ship with the app for anything that has to look identical elsewhere.

Exporting a picture says it needs a browser. PNG, JPEG and the other pictures are rendered by a Chromium-based browser. Every supported version of Windows has one. From the command line, set BINSE_BROWSER if it is somewhere unusual.

The palette is missing its icons. In a source checkout, run node web/scripts/gen-libraries.mjs: the icon packs are generated rather than committed. In an installed copy this cannot happen.

Routing feels slow on a very large page. Routing is incremental and budgeted, but a page with thousands of connectors is work. Split it across pages, or turn off re-route the lines in the layout settings while you are arranging things.

The assistant is not there. It is Binse Enterprise only, and it needs a licence key that grants Enterprise on this computer as well as the Enterprise installer. Enter the key and it appears without a restart.

Working together says there is no relay. The relay is a server your organisation runs; ask your administrator for its address. Everything else in the app works without it.

Something crashed and I had not saved. The next start offers the recovery copy, written every 30 seconds while there was unsaved work. Turning autosave on keeps the diagram's own file up to date as well, so there is less to recover in the first place.

Autosave has gone off by itself, in amber or in red. Amber means it has nowhere to write: the diagram has never been saved, or it is only a file on this computer. Click the switch, save it into Binse once, and autosave starts. Red means a save failed, and the reason is on the switch: the file may be read-only, or the folder or drive it lives on may have gone. Your work is still here and still unsaved. Sort the problem out and turn the switch back on.

Glossary