Paste from Excalidraw
You don't need a file to bring a drawing across. In Excalidraw, select what you want and copy it (Ctrl+C, or Cmd+C on a Mac), then click into a livediagram tab and paste (Ctrl+V or Cmd+V). The drawing lands where your pointer is, or in the middle of the view, and stays selected so you can move or restyle it at once. One undo takes the whole paste back.
- In Draw mode, strokes become pen strokes, text becomes text boxes that fit their words, and Excalidraw's black ink follows the board: dark on a light board, light on a dark one. Its blue, green, red and orange become the matching pen colours.
- In any other mode, such as Diagram, the drawing comes in as diagram elements, just as importing the file would.
- You can also drag a
.excalidraw file, or a PNG or SVG exported from Excalidraw with Embed scene on, straight onto the canvas.
When something couldn't come across exactly (Excalidraw groups, for example), a small note at the bottom of the canvas lists what changed. Pasting adds to the tab; it never replaces what is already there.
Excalidraw From a Start Screen
On an empty tab's start screen, Import offers Excalidraw beside the other tools. Picking it opens this Import dialog at the Excalidraw format, and the scene lands in the tab you are on. To bring several boards across, open a new canvas (or add a tab) for each.
Images in an imported file
When an imported file carries images, they come across with it. Each one is resized in your browser (to at most 2048 pixels on its longest side) and saved to your image gallery, so it counts towards the gallery's space like any image you add yourself. An image you already have is recognised and reused rather than stored twice.
While the images save, the panel counts them ("Importing images 3 of 12…"). When it finishes, the dialog tells you how many were imported, how many were already in your gallery, and how many were left as placeholders and why: the file didn't include the image, its format couldn't be read, it was too large, or your gallery is full. The rest of the file always imports. Double-click a placeholder to add its image later.
In a canvas kept Offline, images are kept inside the canvas in this browser instead of your gallery, up to a limit for each import.
Importing from draw.io
To bring many draw.io files across at once, each as its own canvas, use Import from draw.io on the Explorer page.
A draw.io file can hold several pages. The first page replaces the active tab, and every further page becomes a new tab right after it, named after its page, so a three-page file arrives as three tabs. One undo takes the whole import back, new tabs included, and links between pages become links between the tabs.
What carries over:
- Shapes keep their kind, size, position, colours, borders, rounded corners, rotation, opacity and shadow. Flowchart shapes (process, decision, terminator, document, database, data) map one to one; less common ones come in as the nearest livediagram shape.
- Text keeps its formatting: bold, italic, underline, colours, links, headings and lists in rich labels, plus font size and alignment.
- Connections follow the route draw.io itself draws: the same corners, curves and shared trunks, attached to the shapes (or other connections) they join at the nearest connection point, with their arrowheads, line style and labels where draw.io puts them.
- Swimlanes become lanes, with an upright title down the side where draw.io writes it upright, UML classes and ER entities become records with one row per field, tables become tables, and draw.io layers become the tab's layers, hidden and locked ones included.
- BPMN shapes (events, tasks, gateways, data objects and data stores) become livediagram's BPMN elements, markers included, with an event's or gateway's name as text under it.
- Cloud and network icons (AWS, Azure, Kubernetes, servers, users and more) become the matching livediagram icon where there is a clear match. A stencil with no match comes in as a labelled box, so nothing disappears.
- Colours left at draw.io's default take on your tab's theme, like anything drawn in livediagram, and so do near-black text and near-white panels, which draw.io's own dark mode turns too. Other colours you picked in draw.io stay as they were.
- Text keeps its room. livediagram's text is a little larger than draw.io's, so the whole page is scaled up to match, and every label fits its shape as it did in draw.io.
Pictures embedded in the diagram come across like any imported board's images: resized and saved to your gallery. Pictures linked from the web, or from draw.io's own icon library, are never fetched; they come in as placeholders.
When anything could not come across exactly, the dialog shows one summary before closing, the same one every import uses: how the images came across, then one line per kind of change with its count: shapes that came in as labelled boxes (with the stencil names), arrowheads livediagram draws differently, labels shortened to fit their shapes, hidden items left out, and so on.
Importing from Visio or Lucidchart
The Visio card takes a .vsdx file, the drawing Visio 2013 and later saves. A .vsdx is a file rather than text, so there is no box to paste into: drop the file on the panel, or press Choose a File.
From Lucidchart, open the diagram and choose File, Export, Visio (VSDX), then import the file it downloads. Lucidchart has no file of its own that other tools can read, so its Visio export is the way across. An older binary .vsd, or a Visio 2010 .vdx, is turned away with a note: open it in Visio and save it as a .vsdx first.
Like a draw.io file, the first page replaces the active tab and every further page becomes a new tab named after its page, and one undo takes the whole import back.
What carries over:
- Common shapes from Visio's Basic Shapes and Basic Flowchart Shapes stencils map to livediagram's own: rectangles and processes, rounded rectangles, ellipses and circles, diamonds and decisions, start and end, documents, data, databases, hexagons and preparation, triangles, trapezoids, parallelograms and stars. Less common flowchart shapes (predefined process, stored data, manual input, delay and the like) come in as the nearest livediagram shape.
- Shapes drawn without a stencil, which is how Lucidchart's export writes most of its shapes, are read from their outline: boxes, rounded boxes, ellipses and diamonds come in as those shapes.
- Every other shape comes in as a text box holding its words, so nothing you wrote is lost. A shape with no words of its own shows its stencil's name instead, such as "Server".
- Text keeps its size, colour, bold, italic, underline and alignment, and shapes keep their position, size, fill, line colour, line style and rotation. Colours left at Visio's defaults take on your tab's theme.
- Connectors stay attached to the shapes they are glued to, bend where Visio bends them, and keep their arrowheads and labels. A loose connector end stays where it was.
- Pictures in the drawing come across like any import's images. A picture Visio stores in a format livediagram can't read (an EMF, or an embedded document) comes in as a placeholder.
Visio's layers, shape data, hyperlinks, comments, gradients and shadows don't come across. When anything changed on the way in, the dialog shows the same summary every import uses, including how many shapes came in as text boxes and which stencils they were.
To bring many Visio files across at once, each as its own canvas, use Import from draw.io on the Explorer page: it takes .vsdx files too.
Importing a BPMN file
The BPMN card reads BPMN 2.0 XML, the .bpmn file Camunda Modeler, bpmn.io, Signavio and most process tools save. Paste the XML or pick the file (.bpmn or .xml). It replaces the tab in one undo step, like every import.
- Every shape lands where the modeler drew it, at its size, as a BPMN element: start, intermediate and end events with their message, timer, error and terminate markers, tasks with their user, service, script, manual, send, receive and business rule markers, collapsed sub-processes, gateways, data objects and data stores.
- Pools and lanes become lanes, their titles upright down the left side.
- Flows stay attached to the shapes they join and bend where the modeler bent them: sequence flows as arrows with their names, message flows dashed with an open circle and an open arrowhead, associations dotted.
- Names of events, gateways and data sit under the symbol as their own text, where the modeler put them; a task's name is inside it. Text annotations become text, and groups become dashed frames.
- Colours set in the modeler (BPMN in Color) carry over.
Only the file's first diagram is drawn. Anything livediagram has no symbol for yet (a signal or escalation event, a choreography) comes in as the nearest symbol or a labelled box, and the dialog's summary says what changed. A file over 5 MB, or with no diagram layout saved in it, is refused with a message saying why.
Drawing your infrastructure
The Infrastructure card reads the files that describe what you deploy and draws them as a laid-out architecture diagram, each box wearing its technology's icon. Drop several files on the panel at once, or press it to choose them, and each becomes a row you can remove; there is also a box to paste one more file into. livediagram tells the formats apart by their names, or by what they hold, and draws every file into one diagram.
- A
wrangler.toml becomes the Worker, with a box for each binding joined to it: KV namespaces, R2 buckets, D1 databases, queues (producers and consumers), Durable Objects, service bindings to other Workers, Workers AI, Images, static assets, Vectorize indexes, Hyperdrive, Analytics Engine datasets, Workflows and Browser Rendering. Routes and cron triggers point into the Worker.
- A Compose file becomes a box for each service, its icon picked from its image (Postgres, MySQL, Redis, MongoDB, Kafka, NGINX, RabbitMQ, Elasticsearch and more, Docker for the rest), with its image and ports in its note.
depends_on and links become arrows, named volumes become cylinders, and networks frame their services (or become hexagons when a service sits on two).
- A Terraform file (
.tf) becomes a box for each resource and module, its icon picked from the resource type (AWS, Azure and Cloudflare resources, Terraform for the rest), with the type in its note. A resource that names another (aws_iam_role.lambda.arn, depends_on) gets an arrow to it, and a joining resource such as a policy attachment becomes a dashed line between the two it joins. Variables, outputs, providers and data sources aren't drawn.
- A Kubernetes manifest becomes a box for each Deployment, StatefulSet, DaemonSet, ReplicaSet, Job, CronJob and Pod, each Service, Ingress and PersistentVolumeClaim. A Service points at the workloads its selector picks, an Ingress at the Services it routes to, and a workload at the claims it mounts (dashed). Objects in two or more namespaces are framed by namespace.
- Big systems stay readable. Past 40 boxes, services that share a name prefix (
billing-api, billing-worker) are framed together, and past 150 the rest fold into one last box that lists them.
When a wrangler.toml holds [env.…] tables, a Wrangler environment choice appears: Top level (the default) or any environment the files name. It applies to each file that has it; a file without it draws its top level, and the summary says so.
Keys that aren't drawn, such as vars or observability, are never an error: once the tab is drawn, the dialog lists them in one line so you know what was left out. A file that can't be read says why, naming the file and the line to look at, and your files and paste stay in the panel to fix.
You can also draw straight from your repository on the command line: livediagram link from-code reads every wrangler.toml, Compose and .tf file you name (or those in the folder) into one tab.
Updating a diagram from its files
A tab drawn by the Infrastructure card remembers the files it came from. Open Import on that tab again and a From Its Files card sits above the formats, naming the files and when they were read. Press Update From Files to read new versions into the same tab:
- Each remembered file is a row saying Kept (its text is used again), Replaced (you chose or pasted a new version; "No changes" when it reads the same), Needed (it was too large to remember, so choose it again or remove it) or Removed. Each row can Choose a new version, Paste one, or Remove it (and Keep it again). Drop new versions on the list and each replaces the file of the same name; Add Files brings in files the tab didn't come from.
- Update Diagram keeps your work. Boxes stay where you put them, at their size, in their colours, and a label, note or icon you changed stays yours. A box new in the files is placed beside the box it joins, clear of everything else, and its frame grows to hold it. A box the files no longer name is removed with its lines. A box you deleted stays deleted, and anything you added yourself is never touched.
- It is one undo step, and the summary says how many boxes and lines were added, removed and kept, and how many of your edits were kept.
On the command line, livediagram link from-code --doc <document> --tab <tab> --update does the same from the files in your repository, and prints the command for each tab it makes. In a repository linked with livediagram.toml, list each tab's files under [[sources]] and a plain livediagram link from-code updates them all.
Any other import onto the tab (Mermaid, JSON and the rest) replaces it with something no file drew, so the tab forgets its files. Updating is for Editors; it isn't offered on a locked tab, and from the command line it needs edit access.