Changelog¶

All notable changes to Osmosis will be documented in this page.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.0.2] - 2026-07-25¶

Added¶

Notes as Flashcards (Line Cards)¶

• Generate flashcards from note command (palette + file menu) tags eligible lines with native ^os- block IDs behind a confirmation modal; re-runs are incremental
• Every tagged line is its own FSRS card — front is the ancestor breadcrumb (plus configurable preceding-sibling context in sequential study), back is the line
• Line-card schedules stored in osmosis-schedule note frontmatter (atomic processFrontMatter writes, debounced and flushed at session end)
• Line cards count in decks and sequential study, with per-note (osmosis-line-cards: false) and global opt-out
• Deleted block IDs soft-delete their cards; history re-links when the ID reappears

Study Modes¶

• Reading view Peek and Study header actions: hide all line-card lines for casual review, or study due/new lines top-down with an inline rating bubble, progress pill, and completion toast
• Spatial study reworked: only due or new line-card nodes hide behind ? placeholders (the map stays expanded), tap to reveal, rate via a bubble anchored to the node (keys 1–4), progress pill + Stop, completion toast
• Spatial peek mode on the mind map — hide all line-card nodes, reveal freely, nothing recorded
• Study this branch now scopes spatial study to a subtree's due cards
• Transcluded notes are studiable on the host map: their nodes hide/reveal in spatial study and peek, and ratings write to the embedded note's own osmosis-schedule

Line Card Control¶

• Add / Remove line cards by editor selection or mind map node — editor commands, editor context menu, and node context menu. Removing an ID you authored yourself is confirmed behind a link-break warning
• Exclude from study / Include in study pauses a card without losing its FSRS history: skipped by peek, study, the sequential queue, and dashboard counts. Stored as disabled: true on the card's osmosis-schedule entry; rating never clears the flag, and re-including never clears the schedule
• The sequential study modal's exclude button now routes line cards through the schedule store (fence cards keep their exclude: metadata)
• Multi-line prose runs are split into one block — and one card — per line, matching the mind map's save convention

Cloze Syntax¶

• Unified cloze grammar across prose and code. :::text::: joins ==text== and **text** as a third prose delimiter — its markers are stripped from both sides, leaving no visual residue
• c<N>: grouping works on any delimiter (==c1:Paris==, **c1:Paris**, :::c1:Paris:::). Occurrences sharing a number blank and reveal together as a single card, including across the prose/code boundary inside one fence
• Inline code clozes — :::c<N>:token::: blanks individual tokens inside a line of code, instead of whole lines only
• Line- and region-level code markers gained the parallel -c<N> suffix: # osmosis-cloze-c1, # osmosis-cloze-start-c1 / # osmosis-cloze-end-c1
• Group numbers you write are preserved verbatim on card IDs, so adding a cloze later never renumbers existing schedules. Unlabeled clozes are numbered above the highest labeled one, in source order

Study Sessions¶

• Intra-session learning and relearning steps. Cards rated Again, or still working through their learning steps, reappear within the same session after a timer-based delay, matching Anki's behavior. When only pending timers remain, a "Waiting for next card" countdown appears
• New Learning steps and Relearning steps settings (e.g. 1m, 10m); step position persists in the fence's scheduling metadata
• Multi-level undo in sequential and contextual study via Ctrl+Z or the undo button — reverts ratings (restoring the previous FSRS schedule) as well as excludes
• Exclude card toggle in both sequential (E) and contextual study, written as exclude: fence metadata, with eye / eye-off icons
• Go to card (G) opens the source note and scrolls to the card's line
• Contextual study keeps the answer visible after you rate it, rather than collapsing it away

Study Modal¶

• Action icons (open note, exclude card, undo) moved into the modal's top-left header strip, aligned with Obsidian's close button as round 42 px buttons
• Optional deck breadcrumb showing the deck of the card currently on screen — Settings > Osmosis > Show deck breadcrumb in study modal, off by default
• Cloze reveal now replaces the front in place instead of stacking the back beneath it, so your eye stays anchored on the same text and only the blanks fill in. Basic front/back cards keep the stacked layout, and live preview still shows both sides for editing
• Obsidian-style language labels on code blocks in sequential and contextual study cards
• Rating buttons share equal widths across the bar, are taller on desktop with softer 8 px corners, and the flip button is pill-shaped; Show answer matches the rating buttons' height

Dashboard¶

• Decks auto-indent to mirror your folder hierarchy. Cards without an explicit deck use their full folder path (Study/Math/Algebra) rather than just the parent folder name, and intermediate folders holding no cards are pruned so only meaningful levels appear. Explicit deck assignments keep their full slash-separated hierarchy without pruning

Mind Mapping¶

• Reading mode for the Mind Map View, mirroring the Markdown view's reading toggle. The map is safe to explore but cannot be mutated: every mutation path is guarded (keyboard, ribbon, context menu, double-click), node drags pan the viewport instead, resize handles hide, the ribbon collapses to navigation actions, and context menus keep only Copy / fold / Study this branch. Spatial study and peek stay fully available — ratings are study metadata, not map edits
• Toggle reading mode via the header action, Ctrl+E, or the Toggle mind map reading mode command. Mode persists per pane; the new Default mind map mode setting (Editing / Reading / Reading on mobile only) decides how new maps open
• Callouts and blockquotes are one node: consecutive >-prefixed lines (callout title + body, multi-paragraph quotes, callouts containing lists) collapse into a single node and a single line card, the way code blocks and tables already did. Stacked callouts stay separate
• Expand transclusions setting (default on): embedded branches load expanded when a map opens; turn off to restore lazy collapsed loading
• Embedding the same note twice no longer confuses the map — each embed instance gets independent nodes (fixes selection jumping to the other copy and study/peek hiding only one)

Mind Map Undo / Redo¶

• Mind map edits now have self-contained undo/redo. Map writes bypass CodeMirror, so Ctrl+Z previously did nothing on a standalone map; every mutation is now recorded as a file snapshot with one undo per map operation
• Map-level style changes (theme, layout, direction, balance, side, background, branch line, default fill/border/text/shape/width, and the reset buttons) undo alongside node-level changes, and Ctrl+Z works while the properties sidebar holds focus
• A color-picker drag coalesces into a single undo step instead of dozens
• New Undo history settings group: Undo steps (default 50) and Undo memory cap (MB) (default 20). Whichever limit is reached first drops the oldest edits

Styling¶

• Block-ID style selectors (^os-a1b2c3) in osmosis-styles, preferred over stable IDs and tree paths; the format panel writes them automatically and migrates legacy _n: entries

Documentation¶

• Full documentation site covering mind mapping, flashcards, line cards, and study modes, with a rebuilt homepage: a single hero and call to action, four story sections carrying one idea each, and a compact feature grid. Persona and comparison copy moved to a dedicated Use Cases page
• Link previews (Open Graph and Twitter card meta) now read their description from the site configuration, so the tagline can't drift out of sync

Changed¶

• Minimum Obsidian version is now 1.13.0 (was 1.10.0). Osmosis adopts the declarative settings API introduced in 1.13, which makes every Osmosis setting discoverable from Obsidian's settings search
• Daily new/review card limits are proper number fields with inline validation instead of free text
• Properties sidebar dropdowns use sentence case — (None) and (Inherit)
• Card ID suffixes collapse into a single c<N> namespace; the separate i<N> namespace used by inline code clozes is gone

Fixed¶

Cloze & Scheduling¶

• Cloze-only fences silently lost review progress. Fences producing only derived IDs (abc-c1) never had an id: line written, so the writer couldn't find the fence on review. Three interlocking causes, all fixed: the -r / -c<N> suffix is stripped when choosing what to inject; only known keys (id, exclude, bidi, type-in, deck, hint, and schedule fields) count as metadata, so a prose line shaped like word: value is no longer swallowed; and an injected id: adjacent to content gets a separating blank line
• Inline cloze cards couldn't have their schedules written at all — the ID and key regexes only matched -r and -c<N> suffixes, so the write failed silently
• :::prose::: clozes weren't treated as cards in reading and editing view; the contextual renderer's regex only matched == and **
• Grouped inline clozes left the c1: label visible on the rendered answer (the renderer matched \d+: instead of c\d+:)

Study¶

• Flashcards didn't render in reading view on first paint — the deferred render fired before the element was attached, so reading-view detection returned null. Now driven by requestAnimationFrame
• A deferred learning card whose timer fired while another card was on screen was spliced in at the visible card's index, pushing it down one slot — flipping then revealed the wrong card's answer
• Obsidian's built-in code-block copy button no longer appears on study cards

Mind Map & Plugin¶

• Structural edits across a transclusion boundary no longer corrupt the map. Moving, copying, or re-indenting a line keeps its trailing block ID (and with it, its card history); a node containing an ![[embed]] moves, copies, and deletes as one unit instead of orphaning the ![[…]] line; drop targets use host-file coordinates so an edit can never splice one file's offsets into another's bytes; and an edit, undo, or redo that carried an embed keeps it rendered instead of collapsing to a bare placeholder
• Reading-view study and peek hide a callout that contains a list as one block, matching the mind map (previously the two surfaces disagreed)
• Both-sides balance: the secondary group is reflected about the pivot's center, removing an extra pivot-width gap on the left side
• Properties sidebar rows render correctly on newer Obsidian layouts. Obsidian now draws each setting row as a rounded card with the description stacked differently, which left the card edge touching the label and control and wrapped descriptions raggedly. Rows are laid out explicitly instead of inheriting Obsidian's defaults: name and control share the top line, the description spans the full width beneath, each row carries its own inset, and sliders are pinned narrow so they neither overflow the section nor crowd out the label
• versions.json shipped a template placeholder ({"0.1.0": "0.15.0"}) for a version that never existed, so Obsidian could serve the wrong build to older apps
• Resolved the findings from Obsidian's Community Plugins review: 117 ESLint errors to zero, frontmatter callbacks and undocumented Obsidian internals properly typed, createEl/createDiv over raw DOM calls, window-scoped timers for popout windows, instanceOf() for cross-window checks, 11 deprecated setDynamicTooltip() calls removed, 13 redundant type assertions dropped, and corrected repository links in the README
• An ordered list's number was read through an unchecked as number cast on untyped node metadata; it is now narrowed properly, so a malformed value can't reach the renderer as a number

[0.0.1] - 2026-03-11¶

Initial release of Osmosis — an Obsidian plugin that unifies mind mapping, flashcards, and spaced repetition into a single learning system.

Added¶

Mind Map View¶

• Interactive SVG mind map generated directly from Markdown structure (headings, lists, bullets)
• Bidirectional sync between the Markdown editor and mind map view
• Pan (scroll) and zoom (Ctrl+Scroll Lock) navigation
• Collapse/expand nodes and keyboard navigation (arrow keys, Tab, Enter)
• Multi-node selection with ++shift+click++
• Drag-and-drop node repositioning with live Markdown updates
• Cursor sync between editor and mind map
• Viewport culling for large maps (1000+ nodes)
• Direction-aware arrow keys for indent/outdent
• Right-click context menu on nodes
• Mind map / note view toggle from icons and file menu
• Pin/lock toggle to prevent accidental edits
• Toolbar with add child, insert parent, add sibling, move up/down, delete, refresh, and properties buttons

Node Rendering¶

• Rich content rendering: images, audio, video, YouTube/Vimeo embeds, Excalidraw drawings
• Code block rendering with proper tab sizing
• Interactive checkbox toggling
• Bullets, links, tables, and ordered list renumbering
• Inline editing via embedded Obsidian editor (F2 / double-click)

Transclusion¶

• ![[linked-note]] and ![](path) embedded as sub-branches in the mind map
• Recursive expansion with cycle detection
• Lazy loading for large transclusion trees
• Visual indicator for transcluded branches
• Edit propagation — changes to transcluded nodes write back to the source file

Styling & Themes¶

• 12 preset themes with accent-colored UI
• LCVRT cascade system (Layout, Class, Variant, Role, Theme) for style resolution
• Per-node style overrides stored in osmosis-styles frontmatter
• Heading-level typography (H1–H6 sizing)
• Topic shapes (rectangle, rounded, ellipse, etc.) with inscribed content fitting
• Style classes with local and global scope (create, save, rename, delete)
• Style variants for quick style switching
• Copy/paste node styles
• Branch line patterns (solid, dashed, dotted) and taper modes (none, fade, grow)
• Font family picker with system font discovery and WOFF2 drop-in support
• Drag-to-resize nodes
• Color picker and alignment controls
• Properties sidebar with per-map settings and tabbed layout
• Map-level global styling controls and custom theme creation
• Reset all styles with confirmation dialog

Flashcards & Spaced Repetition¶

• FSRS scheduler for spaced repetition
• Card types: heading cards, cloze deletions (bold syntax), code cloze (comment annotations), and explicit front/back fences
• Note opt-in system for card generation (per-note, tag-based, folder-based)
• Stable card identity across edits with orphan detection and conflict resolution
• Inline scheduling data stored in code fences (no HTML comment clutter)
• Deck assignments via frontmatter

Study Modes¶

• Sequential study: Classic Anki-style card-by-card review modal with rating buttons
• Spatial study: Study cards directly on the mind map with flip-to-reveal
• Contextual study: Study inline within the note view
• Dashboard with deck overview and study statistics
• Scrollable study cards so rating buttons don't overlap answers
• Mobile-friendly study UI

General¶

• Works on both desktop and mobile (touch support)
• Undo/redo support for style changes and node operations
• Per-map settings stored in frontmatter
• GPL-3.0 license