書書き庭KAKINIWA
STATUS

What it can and can't do yet

This is the status of the public beta. This page is updated from time to time. (Last updated: 2026-11-09 · Latest released version: v0.14.0)

For an introduction to the features, see the features page. This page is a detailed list of what is supported, plus caveats, known issues and future plans.

✅ What it can do now

  • Script notation — Just write “Name: line” and you have a script. Expressions go as “Name (smile): line”, an SFX inside a line as [SE:name], and a pause as [pause:0.5]. Within one episode you can divide things with “# Title → ## Scene” headings, and notation such as backgrounds, SFX and ruby can be inserted on the spot with a single Alt+letter keystroke
  • Direction commands (@ at the start of a line) — Besides @bg / @BGM / @bgmstop / @SE, you can set positions with “@place Sora:-25 Hina:25,10(80)”, switch portraits with “@portrait Sora:uniform”, add props with “@prop Cup:-40,18(18)”, and add foreground effects with “@fx focus lines:10,-5(80)@1.5”. Text windows use “@window Novel frame”. To swap props in and out of the same spot, set the spot first with “@slot”, then swap with “@prop Fig1:frame name” (to remove, use @propclear / @window 消去)
  • Four display modes — Cards / Focus / Source / Novel. Also ruby, emphasis dots and tate-chu-yoko (horizontal text within vertical text)
  • Interface language — Choose from Japanese, English and Spanish (switch on the top screen or in Settings). You can also write script notation using the spelling of your display language. Files are saved with English spellings, and older scripts written with Japanese spellings still open as they are
  • Writing support (proofreading) — Detects unregistered characters, inconsistent first-person pronouns and typos, with one-click fixes. v0.11.0 added five more kinds: inconsistent proper nouns (spellings that differ from character names, world entries and the readings dictionary only in script type, width or case), forms of address (lines that differ from the form of address set on a line of the relationship map), sentence endings (lines that don't match the “speech style / endings” in the profile), your own proofreading rules (register a pattern to find → a reason; as plain text or a regular expression, two levels of warning or notice, shared across all projects), and sending a finding to Tasks with one click. Character and world settings appear next to the script, and you can edit them on the spot
  • Character settings — First person, speech style, banned words, expression tags, and changes over time / timeline. Portraits support PSD plus an expression editor (tabs let you edit the expressions of several portraits one after another)
  • Reference materials — Relationship Map (Shift+drag to draw a line; click to edit its name, color, thickness, line style and arrows. Multiple maps and PNG export. Since v0.11.0, lines can have “from when / until when”, and the “As of” control on the toolbar shows only the relationships of that period. Since v0.12.0, ◀ ▶ step through one episode at a time and summarize the relationships that appeared or disappeared in that episode. You can also drag to move the display position), World & Terms, Plot (a four-act kanban. Since v0.12.0, cards can be tied to scenes in the script and jump straight to them), and Pacing (target and estimated length per episode, plus a “Whole project” view that bundles all episodes into total length, differences and per-episode allocation. Settings left behind after a heading was rewritten are shown as well)
  • World links, aliases and maps (v0.13.0) — Write [[entry name]] in the text of a world entry and it becomes a link to that entry, with the link target and “referenced by” shown side by side. Entries can also be looked up by their “aliases” (nicknames, former names). You can also turn an asset image into a map and tie world entries to it with pins
  • In-story timeline and character ages (v0.13.0) — The Plot now has a timeline ordered by numeric years of the story's calendar. From a character's birth year and an episode's in-story year, the character page also shows their age. The Dashboard also shows “Today's characters”
  • Custom-font text (fictional scripts, etc.) (v0.13.0) — If you set an asset font to “Also use for script text”, the script and Vertical Preview are drawn in that font without installing it on your computer. You can choose the range it applies to (only the fictional-script parts, everything, or a range of characters) and turn the display on and off. Video subtitles (YMM4) are not covered
  • Novel mode (v0.14.0) — “Novel” (Ctrl+4) is now a display mode for scripts. Write a whole episode as one continuous manuscript. Indentation is added on screen only; no spaces are written into the file. You can fold chapter and scene headings, ruby and emphasis dots are shown, and you can write vertically (digits and Latin letters stand upright one character at a time). It also supports exporting to an e-book (EPUB) and proofreading typos in narration and stage directions
  • Character table for fictional languages (v0.14.0) — The world-building “Language” now has a character table. You can give names and meanings to the characters in the font and the ones used in your work. From the table you can export a user dictionary for your input method (Google Japanese Input, Microsoft IME) and a decoding chart for readers (CSV, HTML). From a font that has glyphs in the Latin-letter positions, you can also make a new font reassigned to fictional characters (the original font is not changed)
  • Asset library — A tree view that shows types and folders together. You can create organizing folders on the spot and put one asset into several folders. Reference-location mode (use files without moving them from their original folder), fill rate, sort order (by name, most used, newest, manual) and pinning. It stays light even with thousands of items
  • Idea Notebook — Switch ideas with one click between “this project only” and “shared across all projects”. For dialogue ideas, pick the speaker and the scene and add them straight to the script
  • Importing script files — Load txt / md / csv / tsv directly. Formats such as “Name「line」” are tidied automatically, and for csv and tsv the “speaker, line” columns become the script (older Notepad character encodings are fine too)
  • A different portrait for just one line — Write “portrait name/expression” in the parentheses of a line to switch the outfit or look for that line only (example: Sora(casual/smile): I'm home). Choosing from the expression menu writes it in automatically, so you don't have to type it by hand
  • Right-click menus — Right-click works almost everywhere in the app: script lines, characters, plot, assets, the project list, and more (duplicate, move, delete, add and so on). In text fields it gives you cut, copy and paste. Drag-to-reorder works too: just drop roughly where you want it
  • Splitting and joining lines — Press Enter in the middle of a line to split it in two, and press Backspace at the start of a line to join it to the previous line by the same speaker
  • Line breaks inside a line — Write [br] in the middle of a line (enter it with Ctrl+Enter or the “Line break” button on the symbol bar) and the subtitle, preview and YMM4 text break into two lines there. The script file stays one line = one line of dialogue. Voice reading continues without pausing at the break. Line breaks are kept when you paste dialogue written on two lines or import CSV/TSV
  • Official community (Discord) — An official server run by volunteer users. Besides asking for help and swapping tips, you can just start a thread in the “Questions & help”, “Feedback & requests” or “Bug reports” forums and it will be received (we organize the contents, and ask right there if anything is missing). Join here. If you don't use Discord, please use the chat intake on the Contact page
  • Vertical Preview — Switch between novel, script and publication layouts. Print or save as PDF as it is
  • Voice preview — Listen as you write, line by line or straight through. Supported engines: VOICEVOX, VOICEVOX-compatible engines (AivisSpeech, COEIROINK, SHAREVOX and others; add as many connection targets as you like and use them together), CeVIO AI, the Windows built-in voice (SAPI5, Windows only), and Yukkuri voice (AquesTalk / AquesTalk10. It is not bundled with Kakiniwa, so get it from AQUEST's page and choose the downloaded zip in “Quick setup” in Settings. To have lines containing kanji read aloud, install the reading engine and dictionary with the “Install” button for the “Reading engine” in Settings (not needed for lines in hiragana and katakana only. After installing, you can update it to the latest with a single button, separately from Kakiniwa updates). English words not in the dictionary are read letter by letter. For commercial use, you need a usage license from AQUEST). Per-character “voice adjustments” such as speed, pitch and intonation with “▶ Preview”, and the option to turn off engines you don't use. While connected remotely, sound synthesized on the host PC is sent to you, so you can check in the characters' voices even from a smartphone (the voices that arrive are VOICEVOX-type voices; CeVIO AI and Yukkuri voice can't be used on the remote side)
  • Export — YMM4 script (txt/csv/tsv), YMM4 handoff pack, direction timetable, SRT subtitles (can be split by speaker), asset manifest, credits, Relationship Map (PNG), reference materials (Markdown), settings data (YAML), and custom formats. Since v0.13.0 you can also make an e-book (EPUB, vertical text supported) and a viewer page (ZIP, no network needed) that can be read just by opening it in a browser. For film and live action, you can also output a screenplay (text), per-role sides (extracted scripts) and a shooting schedule CSV (v0.10.0). Since v0.11.0, you can list the lines changed since the previous export by line number (a guide to where to re-place things in YMM4), and also produce a recording memo to hand to the people you ask for voices (number of lines, character count, estimated length and readings to watch for, per role). Each character's speaking speed is reflected in the estimated length, subtitles and the handoff to YMM4
  • Dedicated plugin (beta) — In addition to lines, voices, portraits, expressions, positions, backgrounds, BGM and SFX, it places props, effects (image/video) and text windows on the YMM4 timeline all at once. For characters with several PSDs, such as outfit variants, you can assign a YMM4 character to each portrait when importing (the voice and expression for lines in between also go to that character)
  • Bug reports are now easier to investigate (v0.9.2) — When you press “Diagnostics → Send anonymously” in Settings, a report number (#number) appears. If you mention that number on Discord or in a contact message, it is automatically matched against records of your environment and actions at the time, so we reach the cause faster (the contents of your manuscript are not sent)
  • Data — Saved as Markdown + YAML (opens in Notepad and Obsidian; Git-friendly). Version history, autosave and trash are built in

⚠️ What it can't do yet and notes on use

  • Both the app and the dedicated plugin are in public beta — They've reached the point where you can use them for everyday production, but specifications may change. Please send bugs and requests to the community (Discord) or Contact. We'll fix things as fast as we can
  • Expression names can't contain a slash (/) — Since v0.7.0, what's inside the parentheses of a line is read as “portrait name/expression name” (example: Sora(casual/smile)). In scripts that put a / in the expression name itself before that, that line now means something different. Proofreading will point it out; when you find one, change the expression name or rewrite it as a switch with @portrait
  • The dedicated plugin may stop working after a YMM4 update — Tested with: YMM4 v4.54, v4.55 and v4.56.1 / plugin v0.6.0
  • Timeline placement only works as a fresh, all-at-once placement — After you edit the script, you can't apply just the changes (you have to place everything again)
  • Voice generation requires speaker assignment — On the YMM4 side, assign the pack's speakers to YMM4 characters
  • You can't specify a YMM4 item template from the script
  • The official Remote is currently open for free — It does require your home PC (the host PC) to be running. The Cloud edition isn't available yet (in preparation)
  • The Mac version doesn't support automatic updates — Because the app isn't signed, an automatically installed update would stop macOS from opening it. We'll announce new versions, so please reinstall from the official site (the Windows version updates automatically at startup)
  • Yukkuri voice on the Mac version is untested — The code is written for Mac, but we have no development environment for it and haven't been able to check on a real machine (targeting Apple Silicon). If it doesn't play or an error appears, it's set up to show a message that explains the cause, so please send it to us as it is. CeVIO AI is Windows-only
  • Because the app isn't signed, antivirus software may block it — On Windows this tends to happen right after a new version is released, and it clears up within a few days once we file a report. For what to do, see About antivirus false positives. On the Mac version, only the first time, right-click → “Open” to launch it

💬 Things we'll consider depending on requests

  • Support for speech synthesis software other than the above — We'll consider it if requests gather. Tell us which software you'd like to use through Contact
  • Support for editing software other than YMM4 — Likewise, it depends on requests. The contents of the intermediate file (handoff pack) exported for YMM4 are plain JSON, so you're also welcome to read it and build your own integration tool for other software

🔜 Coming next

  • Cloud edition — We're preparing an edition that keeps your projects on our servers, so your PC doesn't need to be on (the Cloud edition is not planned to support voice preview. To listen to voices while you write, please use the Desktop edition)

🤖 About AI integration

  • Notation manual for AI (available now) — We distribute a one-page summary of Kakiniwa's script notation. Paste it into ChatGPT, Claude or similar, and it can write a script in a form you can paste straight into Kakiniwa (page for AI and developers)

Beyond this, there are no features yet that use AI from inside the app. The following is an announcement about our future direction.

Kakiniwa is built on the premise that you can use the AI you want to use. We have the following three options in mind.

  • Operating Kakiniwa from the AI you already use (MCP, etc.)
  • Setting up and using your own API key
  • Kakiniwa provides the AI, ready to use as it is

This page previously said that we would not take the approach of providing the AI ourselves for the third option, but we have changed our policy and put it back on the table for consideration. We don't want anyone to be unable to use it because connecting to an outside AI or preparing an API key is a hassle. If we do provide it, we plan to add it as a new option rather than replace the existing ones.

All three are still at the consideration stage. Nothing concrete has been decided, such as how it would be provided, pricing, which AI would be supported, or how project data would be handled. We'll announce it on this page as soon as it is decided.

🚫 Things we don't plan to do

These aren't "not yet" items. We've decided, as a matter of policy, not to do them.

  • Deliberately making features inconvenient to push you toward paying, and showing ads — features that cost nothing to maintain will stay completely free (how we think about pricing)
  • Saving in a proprietary binary format — your projects will stay human-readable Markdown + YAML. We won't lock them inside the app
  • LAN access (a simplified version of the official Remote) — we had planned it, but dropped it for security reasons (protecting traffic inside a LAN is difficult) and for maintenance reasons. Please use the official Remote from other devices

Kakiniwa's Desktop edition is free, for good. How to get it →

Send bugs and requests through Contact. We fix things as fast as we can.

← Back to top FAQ → Dedicated plugin → Changelog →

If you're curious, just give it a try.