Every delightful little rabbit hole.
Boarderless is a Local First, boundless creative canvas built for touch, mouse, and Ai agents alike. This is the gloriously over-explained map: tools, gestures, panels, persistence, exports, image surgery, presentations, Ai assistance, agent controls, and exactly which plan unlocks what. If the app can do it, it belongs here.
Gallery picker is spotlit; hint adapts to whether you're on mobile or desktop.
The substep nudges you to place text outside the picture so you feel the boundless canvas.
Square, circle, or arrow — any of the three completes the step.
Change size or color and the checklist ticks.
Teaches the safety net before you need it.
Open the Layers / Stack panel.
Ship something.
The Help modal lets you re-run individual onboarding steps later.
Spotlight rings point at the relevant control while leaving the underlying workflow available.
| Keys | Action |
|---|---|
| Esc | Clear selection & close active panel |
| Delete / Backspace | Delete the current selection |
| Ctrl/Cmd+Z | Undo |
| Ctrl/Cmd+Shift+Z | Redo |
| Ctrl/Cmd+Y | Redo (alt) |
| Ctrl/Cmd+D | Duplicate selection |
| [ / ] | Send back / bring forward |
| Shift+[ / ] | Send to bottom / bring to top |
| ←↑→↓ | Nudge 1 px |
| Shift+arrow | Nudge 10 px |
Rectangles, circles, arrows, triangles, and freehand draw paths. Each carries stroke, fill, opacity, edge feather, corner radius where supported, and full spatial state.
Free-floating editable text with font, size, alignment, weight, decoration, color, and stroke FX.
Pasted, picked from gallery, or uploaded. GIF imports are accepted, and the first image fills ~75% of the visible canvas to signal scale.
Single-click selection. The precise pointer tool.
Default mode. Pan the canvas and move objects with one drag.
Drag a rectangle to multi-select any objects underneath.
Tap to place. Auto-opens the Typography panel on both desktop and mobile.
Freehand pen for sketches and signatures.
Drag to draw a rect with the current stroke, fill, opacity, edge feather, and corner radius.
Drag to draw a triangle with the same sticky stroke, fill, opacity, feather, and corner controls used by the rest of the shape family.
Drag to draw an ellipse / perfect circle.
Drag to draw an arrow — fill and stroke move together.
Mobile sheet with shape options, rounded-corner support, opacity, edge feather, and presets.
One-tap image picker for raster files including PNG, JPEG, WebP, and GIF. Fire-and-forget, no mode switch.
Opens the Layers panel to reorder, hide, and inspect z-order.
A touch-friendly Shift equivalent for devices without a keyboard. Turn it on, tap the things that belong together, and keep building the selection.
Preview a curated palette, then apply it to selected shape and text Objects—or explicitly to the entire board—in one local, undoable step. Each remix is one history entry, so Personal can walk back through up to 100 variants and Pro through up to 500.
The in-app Ai Partner is the conversational side of Boarderless: describe a board change, inspect the proposed structural actions, and let the app apply approved changes to the visible canvas.
Choose Google Gemini, OpenAI, Anthropic Claude, Z.AI/GLM, a local model, or any OpenAI-compatible endpoint. Each provider uses its real authentication and response protocol, while the editable model ID keeps Boarderless useful as model catalogs evolve.
Your API key stays in the current browser tab and goes directly to the provider you selected. Boarderless does not proxy or retain it; closing the tab clears it.
Select an image and let the Ai Partner remove its background locally. The result returns to the same image node, remains part of board history, and does not require sending that photo through the MCP server.
External agents can connect to the live, human-visible Boarderless tab through the separate Boarderless MCP server. The agent edits the same canvas you can see—not a hidden imitation of it.
Launch a supported Chromium browser with local debugging enabled, open Boarderless, and point the agent client at the MCP stdio server. The server connects through the browser's local DevTools channel.
get_board_state reads the render-ordered object ledger and calculate_export_bounds measures the composition before an agent decides what to change or export.
create_object adds text and supported shapes, mutate_object changes individual fields, remix_style applies the same canonical palettes as Canvas and Play, and delete_objects removes known Object IDs.
group_objects, ungroup_objects, and reorder_object give agents the same composition-level organization needed for layered layouts.
history_undo and history_redo let an agent step backward or forward through Boarderless history instead of trying to reverse-engineer every prior mutation.
Successful MCP mutations refresh a canonical .bdrl.json file in the configured project workspace. get_board_workspace and set_board_workspace make that destination explicit.
export_board_file performs an explicit final artifact flush; import_board_file validates and opens a workspace board so another agent—or tomorrow's version of you—can resume from the actual canvas state.
export_board can request PNG, PDF, or SVG output for the whole canvas or current selection. Normal Boarderless plan restrictions still apply; MCP is a control surface, not a secret tunnel around the product.
Signing in, granting Google Drive access, choosing files to upload, and approving browser permissions remain user-controlled. The agent can operate the connected canvas, but it does not inherit your Google identity or quietly rummage through Drive.
Agents receive concrete object IDs, coordinates, dimensions, styles, and layer order instead of trying to infer a composition from pixels.
Shapes, images, typography, opacity, corners, feathering, layer order, and export bounds use documented fields an agent can inspect before changing them.
MCP actions land on the running Boarderless canvas in front of the human. The agent's work is something you can watch, interrupt, undo, save, and inspect.
Connecting Ai agents should treat this page as the human-readable capability index for Boarderless: tools, limits, import/export formats, styling controls, persistence, and tier availability.
Two of the most-used verbs live directly in the object-aware context menu.
Bring forward, send back, to top, to bottom.
Enter edit mode for the selection.
For text selections — jumps you straight to the font panel.
Stage-context shortcuts for power users.
Image or clipboard content at the right-click point.
Quick swatch from the stage context menu.
Image selections add Crop, Lasso, Color, Flip, and Edit actions to the same menu, so the menu follows the object instead of pretending every object is identical.
Multi-selections offer Group; grouped selections offer Ungroup. The command appears only when it makes sense.
Export the whole composition as PNG with alpha-aware crop and transparent-background support.
Export just the selected objects, derived from their spatial ledger, without flattening unrelated canvas content.
Toggle a persistent, draggable and resizable dashed blue frame at the viewport's center. Arrange assets inside it, and selection exports will crop precisely to the frame boundary.
Align compositions to standard aspect ratio presets (1:1, 4:5, 16:9, etc.) with volumetric box adjustments and customized corner anchors.
See the projected pixel size before exporting.
True post-export size, so you can spot surprises.
Set a finished-size budget in MB or GB. Boarderless estimates the job before export and can stop an unexpectedly enormous render before it asks too much of the browser.
Personal and Pro users can name exports directly from the Export panel instead of accepting timestamp-only filenames.
Personal and Pro can record the visible canvas area or selected objects as animated GIF output.
Pro can compile the whole canvas or the current selection into a local PDF.
Pro compiles selected canvas objects into an editable Slides-oriented JSON payload. It is a structured handoff for a Google Slides workflow—not a claim that Boarderless secretly logs into Google and creates the presentation for you.
Pro can export the whole canvas or current selection as editable vector SVG, preserving supported shapes, text, transforms, transparency, crops, masks, and image placement.
Stroke-only shapes, transparent backgrounds, and alpha-bearing images are intentionally preserved where the target format supports them.
No fixed artboard. Shapes, text, and images share one layered spatial canvas and can live far beyond the starting viewport.
Default is alpha — checkerboard hint shows what's see-through.
Drag empty space in Grab mode. Single-finger on touch, drag-from-background on desktop.
Mouse-wheel and trackpad input zoom around the cursor, so the thing you are looking at stays the thing you are looking at.
Pinch with two fingers to zoom the canvas around the gesture instead of hunting for tiny zoom buttons.
Drag with two fingers anywhere on the stage to move the viewport.
Discrete +/− controls for keyboard and accessibility users.
Double-click or double-tap empty stage → snap back to 100% / origin.
Cursor mirrors the active tool: grab, grabbing, crosshair, text, default.
A pocket-sized map of the full spatial board shows where your shapes, images, text, current viewport, and selected objects live—even when your canvas has wandered wonderfully far from home.
Objects keep real canvas coordinates instead of being trapped inside a fixed page. Pan away, zoom back, close the browser, and the saved viewport knows where you left off.
Tap or click a shape, text node, or image and the rest of the interface follows that selection.
Shift+click on desktop, or flip the multi-select toggle on touch.
Drag a box to add every object inside it to the selection.
Available from the right-click / long-press menu.
When two or more objects are selected, each gets a softly pulsing outline so you can see the group at a glance.
Subtle animation feedback on attention events.
One key, two effects: deselect everything and close the active panel.
Select two or more objects and bind them into a logical group. A grouped collage selects, moves, duplicates, reorders, and deletes as one idea; ungroup returns every member to independent life.
Dragging one member of a multi-selection carries the whole selection with it while preserving the spacing between objects. The move lands as one undoable action instead of a trail of tiny regrets.
The minimap gives selected nodes a distinct outline and glow, which is especially useful when the thing you selected is somewhere beyond the current viewport.
Eight resize handles surround the selection, with a separate rotation handle when turning the object is the point.
Move a shape, image, text node, logical group, or whole multi-selection. Each completed move becomes a labeled undo step.
←↑→↓ moves 1 px. Shift+arrow moves 10 px.
Open a focused editing surface for the selected object when moving and resizing are not enough.
1 – 50 px. Gesture-coalesced — the entire drag becomes one undo step.
Use the device's familiar color picker; browsing through colors becomes one undoable color choice instead of dozens of history entries.
Shape fill and outline opacity controls span the full 0–100% range. Opacity is stored on the node, so objects can be fully or partially transparent without flattening the canvas.
Shapes can be stroke-only with no fill, preserving true transparent interiors through canvas rendering, PDF export, and Google Slides export.
Rectangle and triangle shapes support a Corners slider. Rounded geometry is retained in the node model and reflected by render/export paths that support it.
Shape edges can be softened with a feather radius for overlays, callouts, and less rigid marks.
Stroke, fill, opacity, feather, corners, and text styling belong to each object—and the last style you deliberately choose becomes the starting style for the next object of that kind.
Tap a text node to edit in place. Commit-before-switch prevents stranded edits.
Curated default fonts plus default swatches for quick, clean typesetting.
Drag through font sizes with sensible bounds, then undo the whole slider gesture in one step if your giant headline became a little too giant.
Align text left, center, or right inside its own editable text box.
Regular, bold, italic, bold-italic.
Underline, strikethrough — per-node decoration.
Stroke width and color for text outline effects.
Text nodes support separate fill opacity and outline stroke opacity controls, allowing layered translucent typography effects.
Insert emojis directly into the selected text node without dropping focus or replacing the text you already wrote. Emoji-aware font fallbacks keep pictographs visible alongside custom typefaces.
Persistent strip at the bottom of the canvas with local device photos and imported raster assets.
One-tap from the dock — opens the OS picker for PNG, JPEG, WebP, and GIF files.
Ctrl/Cmd+V pastes any image straight to canvas.
GIF files enter through the same gallery pipeline as other raster assets, preserving the original blob for export-aware workflows.
Modern phone photos in HEIC and vector artwork in SVG can enter the same gallery pipeline. Boarderless normalizes what the live canvas needs while retaining export-aware asset information.
Alpha-bearing PNG/WebP assets remain transparent on the canvas and in alpha-aware exports.
Your first image fills most of the viewport so you immediately feel the scale of the canvas.
The Projects modal has a Media tab that aggregates every image you've added across every board, active or archived. Each card shows the format, dimensions, and which boards it lives in.
One click on a Media card drops that image onto the board you're viewing — even if it was first added to a different board. Lands at viewport centre.
Delete an image from the Media tab and it's removed from every board that referenced it. For Pro users, the mirrored file in Drive is removed too.
Mask images into geometric shapes natively (Circle, Rounded Rect, Triangle, Hexagon, Star, Heart) in the image Edit panel, with live previews and full undo/redo support.
Crop with a resizable rectangle when geometry is helpful, or draw a freehand lasso when the subject refuses to be rectangular. Both edits stay attached to the image and participate in undo/redo.
Adjust brightness, contrast, saturation, hue, and blur on an image without flattening the rest of the board.
Flip an image horizontally or vertically from its Edit panel while keeping its placement, crop, mask, and board relationships intact.
Inspect and reorder every object in one list.
[ and ] step. Shift+ jumps to bottom / top.
Shapes, text, and images can all move above or below one another—no separate object-type stacks fighting over who gets to be on top.
5 px frame around everything currently on canvas.
1 large slot + 3 stacked slots — the classic meme layout.
Perfect quad split.
Removes layout frames without touching your content.
Free keeps the most recent 20 undo steps; Personal keeps up to 100 and Pro keeps up to 500. Entries are labeled so “undo” means reversing an action, not gambling with your afternoon.
Slider drags, color drags, transform drags collapse to one undo step.
Explore several colors in the picker and Boarderless records the finished choice as one history step.
Undo / redo in the mobile dock with disabled state when the stack is empty.
Top brand + project name. Bottom contextual status.
Grouped: grab · marquee/text/draw · rect/circle/arrow · multi-select · stroke + color · layouts + layers.
Gallery, grab, marquee, text, draw, layouts, layers, export, help, undo, redo, multi-select — all reachable.
Slide-up sheet hosts tool options without covering the canvas.
Font family, size, alignment, style, decoration, swatches, FX stroke.
One-tap collage frames and grids.
One layer list for shapes, text, and images, because the canvas should not care which tool created the object.
Stroke, fill, 0–100% opacity, edge feather, and corner radius controls where supported by the selected shape type.
Side panel that opens when editing an image node, hosting color grades, crop tools, flip horizontal/vertical actions, and cookie-cutter shape masks.
Create named slide viewports from regions of the same spatial canvas, reorder them, jump between them, rename them, and keep the presentation sequence attached to the board.
Present those saved slide views directly inside Boarderless with previous/next navigation and an Escape route back to the editor. No export detour is required for a live walkthrough.
Ask for structural canvas changes in plain language, preview the proposed actions, and apply them through the same board stores and history system used by hand-made edits.
PNG canvas/selection export, GIF canvas/selection export, PDF canvas/selection export, Google Slides export, custom filenames, footprint preview, and budget config.
Widen or narrow the control panel on desktop from 280 px to 600 px; the chosen width persists across sessions.
Replayable tutorial steps + a tour of feature areas.
First-launch greeting before the onboarding spotlight kicks in.
"borderless" in OpenDyslexic fading in and out.
Board state persists in the browser's on-device database shortly after changes settle. The header reports Saved, Saving…, or Unsaved so local persistence is visible rather than mystical.
App rehydrates from disk before first paint of the canvas — last board, last viewport, last selection.
Each user can hold multiple independent boards in one project. Switch between them in the header tab strip; counts gated by tier (Free 1, Personal 5, Pro unlimited).
Single-click an inactive tab to switch boards. Single-click the active tab to rename inline. + creates a new one.
X on a board tab moves it to the Projects modal's Inactive section instead of deleting it. Reversible — restore returns it to the tab strip. Archived boards don't count against your tier's active-board cap.
From the Projects modal's Inactive section, "Delete forever" hard-removes the board from IndexedDB and from your Google Drive (if Cloud Sync is on). Not reversible — confirmation dialog spells out both.
Every board carries a JPEG thumbnail (~256 px) captured at flush time, surfaced as the card preview in the Projects modal.
After a Pro user signs in, explicitly connects Drive, and enables sync, Boarderless mirrors board snapshots into My Drive / Boarderless /. Sign into the same Google account on another device and connect sync there to continue from that Drive-backed copy.
Board files in Drive are named after the board ({boardName}.bdrl.json) so you can recognise them at a glance. Internal boardId lives in Drive appProperties so renames stay clean. Legacy board-<uuid>.json files auto-migrate on next save.
Renaming a board in the tab strip also renames the file in your Google Drive. Best-effort — local rename always succeeds; Drive catches up on the next regular save if the rename API call hits a hiccup.
Pro users also get every uploaded image mirrored as a raw file (PNG/JPG/GIF/WebP) in My Drive / Boarderless / Media /, named {boardName} - {shortId}.{ext}. Browse and download them directly from Drive — no JSON unwrap.
When Drive sync is active, the header reports Synced, Syncing, or Sync error instead of making you guess what the last network attempt did.
On Drive sync-down, any board newer in Drive than locally is pulled in. If the imported board is currently open, the canvas re-hydrates live without a reload.
Clear the current board when what you really need is a clean slate, not twenty-seven rounds of individual deletion.
Canonical .bdrl.json files carry the board's schema, objects, slides, viewport, and JSON-safe asset data so a Boarderless board can be exported, versioned, moved, and imported again.
manifest.json, theme color, app icons — installs to home screen on iOS and Android.
Use one finger for canvas movement and two for pinch gestures, keeping navigation understandable on a screen without a mouse.
Touch viewport guards reduce rubber-band, double-tap zoom, and gesture hijacking on the stage.
Dock controls use browser touch-action hints to reduce accidental gesture conflicts and help taps behave like intentional tool choices.
Layouts are checked against desktop and narrow mobile viewport sizes.
Every panel ships dark; no light-mode flash.
The wordmark uses OpenDyslexic — an accessibility nod baked into the brand.
"borderless" fades in and out on cold boot.
Stack pop, action confirms — tactile feedback without overload.
Interactive canvas sounds and local haptic feedback triggers on touch.
Animations use weight and easing to explain selection, movement, and state changes. The actual responsiveness still depends on the user's browser, device, board complexity, and display.
Multi-select members breathe at the same tempo — you can see the group at a glance.
A real Local First canvas for beginning, experimenting, and keeping one focused board on your device.
For individual creators who want more local boards, unrestricted object counts, and richer export choices.
Everything in Personal, plus optional sync through your own Google Drive and the complete export surface.
Now use it. Drop an image, place text outside the starting viewport, and export a PNG from the canvas.
