A Complexus Lab project. See CONTRIBUTING.md to help.
A local, native mind-mapping app (XMind-style). Tauri 2 + React + TypeScript. Files stay on your machine.
Tabs and menus: the toolbar is three menus (File with templates, recent files and export; Insert; View with structure, theme and panels) plus Undo, Redo, Delete, Find and Save. Each map opens in its own tab (a blank tab is reused; Open… puts a map in a new tab when the current one holds a file or unsaved work); the second row of tabs is the sheets inside one map. Only the first tab keeps the autosave.
Get the latest build for macOS, Windows or Linux from the Releases page. The Windows and Linux builds are not code-signed (on Windows choose More info, then Run anyway). The macOS build is signed and notarized when the release was made with the Apple credentials; otherwise, if the app is reported as damaged, run xattr -cr /Applications/Cartocortex.app once in Terminal, or right-click the app and choose Open. How releases are made: RELEASING.md.
Needs Node 22+ and a Rust toolchain (rustup).
npm install
npm run tauri dev # native window with hot reload
npm run dev # browser only (port 1420; file save/open fall back to download/upload)
npm run tauri build # produces the app bundle
npm test # unit tests for the tree model and layout| Key | Action |
|---|---|
| Tab | New child |
| Enter | New sibling |
| Space / F2 / double-click | Edit text (Enter to commit, Esc to cancel) |
| Arrows | Move selection |
| Delete / Backspace | Delete node |
| ⌘/Ctrl+Z, +Shift | Undo, redo |
| ⌘/Ctrl+S | Save as JSON |
| Drag a node onto another | Re-parent |
| Wheel / drag background | Pan; ⌘/Ctrl+wheel or pinch zooms |
The ‹ chip on a node collapses its branch. The map autosaves to local storage; Save/Open use .json, and Export MD writes a Markdown outline.
src/model.ts– immutable tree ops, tidy-tree layout, Markdown exportsrc/App.tsx– canvas, keyboard handling, undo/redo, save/opensrc-tauri/– Rust shell with the dialog and fs plugins
Tree tables and the unbalanced map, which open as the nearest structure (a branch's own structure is kept only for org charts and tree charts inside a logic chart or map); XMind's smileys, arrows, font sizes and line styles. An import lists what it could not keep.
Fill colour, shape, emoji markers, notes and links (inspector panel); relationship lines and boundaries; Logic and Map (two-sided) structures; five themes with per-branch colours; copy/cut/paste/duplicate (⌘C/X/V/D); find (⌘F, Enter cycles); fit-to-view; .xmind open and export (all sheets; text, notes, links, folding, structure, pictures, relationships, summaries, floating topics, boundaries, icons, fills).
Outline view (toolbar "Outline": Enter = sibling, Tab / Shift+Tab = indent / outdent, shares undo with the map) and PNG / PDF export (the PDF is one page holding an image of the map, so its text isn't selectable).
Summaries: select a topic, click "Summary", then click the last topic of the range (it must share a parent). A bracket and an editable label (double-click) appear beside the span; click the label and press Delete to remove it. They are included in PNG/PDF and .xmind export.
Presentation: "Present" goes full-screen and reveals one top-level branch per step (→ / Space / Enter next, ← back, Esc exit); earlier branches dim.
Multi-select: Shift / ⌘ / Ctrl-click toggles topics; Shift-drag on empty canvas draws a selection box (⌘/Ctrl-drag adds to the current selection); ⌘A selects all. Delete, drag-to-reparent, and the Fill / Shape / Marker controls act on the whole selection, and Summary wraps selected siblings in one click. Copy / paste / duplicate still work on the primary topic only.
Floating topics: "Float" (or double-click empty canvas) adds a free-standing topic with no branch to the root. Drag it anywhere; release over another topic to attach it there. In .xmind export they become ordinary top-level topics (positions are not kept). Deleting topics also removes relationships, boundaries and summaries that pointed at them.
Sheets: tabs under the toolbar, one root topic and one undo history per sheet. "+" adds a sheet, double-click a tab renames it, × then "Delete?" removes it. Save writes every sheet to one .json (older single-map files still open); .xmind import and export carry all sheets. PNG, PDF, Markdown and presentation use the open sheet.
Security: the native window runs under a Content-Security-Policy (src-tauri/tauri.conf.json): scripts only from the app itself, no network connections except Tauri's own bridge, images only from the app, data: and blob:. Inline styles are allowed because topic colours are applied as style attributes. Files you open are parsed as data and never executed.
app-icon.svg is generated by scripts/make-icon.mjs (a side-view brain with a neuron network on a pale teal tile, in the Glossolalie palette: teal #0D9488, navy #1E3A5F and pale tints; the palette and the neuron layout seed are constants at the top). To rebuild: node scripts/make-icon.mjs, render the SVG to app-icon.png at 1024px (any SVG renderer, e.g. @resvg/resvg-js), then npx tauri icon app-icon.png to refresh src-tauri/icons/.
Custom theme: pick "Custom…" in the theme menu to edit the central/accent colour and six branch colours in the panel; changes preview live and commit as one undo step. Colours are stored in the document (themeColors) and only #rrggbb values are accepted.
Look: the UI follows the Glossolalie palette (teal #0D9488 accent, navy #1E3A5F ink, pale teal tints for surfaces, thin 40%-tint rules) with no dark fills and no dark mode; the central topic is a tinted pill with a thicker teal border. The default theme is named "Glossolalie"; the other themes only change branch colours. PNG / PDF export uses the same ink and tints.
Task markers and links: the panel sets a priority (1-5, a numbered disc: darker is more urgent) and a progress (0-100%, a pie that becomes a check at 100%) on the selected topics; both map to XMind's priority-N and task-* markers on .xmind import and export (other XMind markers are dropped). A topic's link shows a 🔗 badge: click it, or use "Open link", to open it in the system browser. Only http, https and mailto links open; a bare host such as example.com gets https://, and javascript: or file: links are refused.
Structures: the toolbar picker offers Logic (right), Logic (left), Map (both sides), Org chart (top-down), Tree (indented list), Fishbone, Timeline, Brace and Matrix. Arrow keys follow the structure (in an org chart, ↓ goes to the first child and ← / → move between topics in a row; in a tree, ↑ / ↓ walk the rows). Summaries are drawn in the Logic and Map structures only. The structure is saved per sheet and maps to XMind's org-chart, tree-chart, logic and map structures; XMind structures not listed open as their nearest match, or as Logic.
Text style: the panel's Text section (and ⌘B / ⌘I) sets bold, italic, strikethrough and a text colour on the selected topics. With the cursor in a topic (or in the notes) the same controls style just the selected words; otherwise they style the whole topic. Notes are now a small rich editor too (plain text underneath, so search still works). Per-word style is saved with the map and shows in PNG / PDF export, but .xmind export writes plain text for titles and notes (topic-level style is still exported). The style is kept in PNG / PDF export and in .xmind files (as XMind's fo:font-weight, fo:font-style, fo:text-decoration and fo:color properties; I have not checked it against a file made by XMind itself). Colours read from files must be plain #rrggbb.
Fishbone: the central topic is the head on the right; top-level topics alternate above and below the spine on slanted bones, with their sub-topics as ribs along each bone and deeper levels hanging to the left of their parent. Arrow keys walk the bones in reading order (← towards the tail, → towards the head). The head is always on the right (XMind's right-headed fishbone).
Old XMind files: .xmind files saved by XMind 8 and earlier (XML inside, content.xml) now open too, with every sheet's title, structure, topics, notes, links, folding and priority / progress markers. Floating topics, relationships, styles and other markers in those files are dropped. Files from newer XMind (JSON inside) are read as before, and take precedence when a file has both. The reader is strict by design: XML with a DOCTYPE is refused.
Timeline: the central topic starts a horizontal line and the top-level topics sit on it, in order. Each topic's sub-topics stack in a column under the first topic, above the second, and so on (nearest first, indented by depth). ← / → move along the line; ↑ / ↓ move into and along a stack. It maps to XMind's horizontal timeline, and any XMind timeline variant opens as this one.
XMind compatibility notes: structure classes, fo:* style keys, marker ids and the style / marker JSON layout were checked against identifiers inside the installed XMind app, but not against a document saved by XMind (none was available on this machine).
More structures: Logic (left) mirrors the logic chart around the root. Brace keeps the logic layout but draws a curly brace around each parent's children (a plain line when there is only a row or two). Matrix (XMind's spreadsheet structure) puts the top-level topics side by side as columns, each with its sub-topics listed beneath. They map to XMind's logic.left, brace.right and spreadsheet.column; XMind's brace.left opens as the right-hand brace, and its row-wise matrix opens as columns.
Saving (desktop app): ⌘S saves to the current file, and asks where the first time (or after New); ⇧⌘S saves under a new name. The format follows the extension: .json (Cartocortex, keeps everything) or .xmind. The window title shows the file name, with a • while there are unsaved changes, and New, Open and closing the window ask before discarding them. "Recent…" lists the last 8 files (kept across restarts). In a browser, Save downloads a .json instead. The map is also autosaved in the app's own storage, which is what you see when you start without opening a file.
.github/workflows/ci.ymltypechecks, builds and tests on every push tomainor afeat/*branch, and compiles the desktop shell on macOS..github/workflows/release.ymlbuilds the app for macOS (Apple silicon and Intel), Windows and Linux when you push a tag such asv0.1.0, and attaches the installers to a draft GitHub release for you to review. The builds are not code-signed, so macOS asks you to right-click the app and choose Open the first time.
MIT, see LICENSE.
Outline formats: Open reads Markdown (.md, headings and nested lists; - [x] task boxes become 100% progress, the first link becomes the topic's link) and OPML (.opml, with notes, links and checked items). "Export OPML" writes one top-level outline per sheet, and Save as can write .opml, .md, .xmind or FreeMind .mm or MindManager .mmap as well as .json.
FreeMind / Freeplane: Open reads FreeMind's .mm maps (plain or gzipped) — topics, notes, links, links between topics, folding, fill and text colours, bold / italic, priority and progress icons — and left-side topics open as a one-sided or two-sided map. "Export FreeMind" (and Save as .mm) writes one root node per sheet, so extra sheets may not show in FreeMind itself; multi-line texts keep their lines, but pictures, relationships, summaries, per-word style, floating positions and structures beyond the sideways layouts are not carried.
MindManager: Open reads MindManager's .mmap maps — a zip holding Document.xml, or the flat XML of the MindManager 10 for Mac builds — with topics, notes, links, task priorities and progress, fill and text colours, folding, relationships (as relationship lines), floating topics and callouts; the growth direction of the central topic opens as Logic (right), Logic (left) or Map. "Export MindManager" (and Save as .mmap) writes one map per sheet, so extra sheets may not show in MindManager itself; topic links to other topics, pictures, per-word style, boundaries and summaries are not carried. The format was taken from Freeplane's MindManager importer and WiseMapping's, and has not been checked against MindManager itself.
Zip-based files (.xmind, .mmap) are refused if they would unpack to more than 64 MB, whatever size they claim, so a booby-trapped map cannot exhaust memory.
Safety rule for opened files: only Cartocortex's own .json keeps its place on disk. Opening a .xmind, .opml, FreeMind, MindManager or Markdown file is an import: the map starts as unsaved, and ⌘S asks for a new file rather than overwriting your original with a lossy copy.
Links between topics: in the panel, "Link to a topic" → "Pick a topic…", then click the target. The source gets a ↪ badge (and "Go ↪" in the panel); clicking it opens any collapsed branches above the target and selects it, also when the target is on another sheet. In .xmind files a topic link is xmind:#<topic id>: it survives a round trip, and XMind allows one link per topic, so an outside web link wins if a topic has both.
Pictures: the panel's "Picture" section puts an image (PNG, JPEG, GIF or WebP) on the selected topic, shown as a small thumbnail in the topic; click the thumbnail for the full image (Esc or a click closes it). The picture is downscaled to at most 800 px and stored inside the map file, so keep an eye on size (the autosave in the app's own storage has a few-megabyte limit). Pictures appear in PNG / PDF export but not in .xmind, OPML or Markdown. Only embedded raster images are ever drawn: SVG and web addresses in a map file are ignored.
Templates: the "Template…" menu next to New starts a map from one of eight starters (brainstorm, project plan as a timeline, SWOT as a matrix, cause and effect as a fishbone, meeting notes as a tree, org chart, weekly plan, pros and cons as a brace map). Like New, it asks before discarding unsaved changes.
Branch structure: select a topic that has children in a logic chart or map and pick Same, Org chart ↓/↑ or Tree →/← in the panel to draw that branch with its own structure, as XMind allows; it round-trips through .xmind.