Writing scripts — the editor and the AI help panel

Writing it yourself is the default. Every AI feature runs only when you click a button (it never steps in on its own), and every suggestion goes through a preview and is applied only after you confirm it.

Editor basics

• Scene and dialogue conventions and autosave are covered in Getting started.
• Focus mode: hides the project tree on the left and the outline on the right at once, leaving only the text.

Toggle it with the ⛶ button at the far right of the top bar (it returns to the same spot). ⌘ . (Ctrl . on Windows) and Esc work too. The setting is remembered, so it stays on next time you come in.

• Collapsing the left tree: use the collapse button at the head of the sidebar to shrink it to an icon rail.

It's expanded by default, and once you collapse it, that choice is kept next time (it doesn't collapse or expand on its own when you move between screens). Even when collapsed, the rail icons take you straight to Project home, Planning, Co-writer and the other menus (hover an icon to see its name), and the episode icon at the bottom opens the episode list beside the rail so you can pick or create an episode. Switching projects and My works are in the expanded sidebar.

• Collapsing the episode list: use the ⌃ button at the right of the Episodes row (the name follows the

project type) to fold and unfold the episode list. It's expanded by default.

• Outline on the right: shows only the scene list for the current episode, in real time. Click a scene to jump to it.

The list shows the scene heading text as is (no separate numbering is added).

• The breadcrumb at the top: "unit · project name · episode name," in that order. The leading unit depends on the project type

(short drama "Episode" · drama "Episode" · film "Chapter" · web novel and animation "Chapter" · novel "Chapter"). Click a name to edit it right there.

• Right side of the toolbar: save status · Production · Comments · Save · the ⋯ menu. The three buttons show both an icon and a label.

When editing isn't available (mobile, no permission), Read-only and the reason appear on the left of the toolbar, and only comments remain in place of the editing tools.

Pick character names from the list

Type @ on a speaker line and the list of characters registered in the project appears. Type part of a name to narrow it, move with ↑↓, and insert with Enter or Tab (Esc closes it). Picking from the list inserts the name exactly and links it to that character.

You can also just type a name that isn't on the list — new characters are registered automatically when you save.

Characters are linked to assets by name. So if you register "Han Jiho" and write "Jiho" in the script, they become two different characters. Picking from the list prevents this mismatch.

A character line with no dialogue yet is saved as is. The empty dialogue spot shows "No dialogue" in faint text — fill it in later. (This text isn't printed in the PDF.)

We let you know when names don't match

If any of the following happens after you save, a one-line notice appears above the editor. It doesn't block your writing, so you can decide later.

• When a similar name has been newly created — "Jiho was registered as a new character — is this the same

character as Han Jiho?" If it's the same person, click [Link to Han Jiho]; if it's someone else, click [It's a new character].

• When two characters share the same name — a newly written line attaches to one of them, and

you're asked which one it is. Use the buttons to link it to the right one.

When linking, the name you wrote in the script stays as is. Only the link is corrected, so your text doesn't change. If the character being merged has images or a definition, they aren't deleted but kept — check them and delete them yourself.

Characters that have already split can also be merged under the character card in Planning ▸ Characters, from "Other characters share this name".

Registering props from the script

Create props in Planning ▸ Props, or drag a name in an action line or dialogue and click [Register prop] to create one right away. Confirm the dragged name once, and a project prop is created, linked to the current scene, and saved immediately. If a prop with the same name already exists, it links the existing prop instead of creating a new one.

• For a linked prop, you can see its episodes and scenes under Appearances on the detail screen, and the link is kept even if you rename it.
• Shape, material, and the reference image are managed in Planning ▸ Props, the same way as character and location assets.
• Even without a reference image yet, the name and definition are passed as text instructions when you bring the scene into production.
• Renaming an episode: click the episode title at the top to edit it in place

(Enter to confirm · Esc to cancel).

• Deleting an episode: [Delete this episode] at the bottom of the toolbar ⋯ menu. The script goes with it and

this can't be undone, so you're asked to confirm once. After deleting, you're taken to the project home.

English screenplays and dialogue in other languages

English screenplays use standard Fountain notation as is (2026-09-24).

• Scene headings: INT. CAFE - DAY / EXT. ROOFTOP - NIGHT. Times of day such as DAY, NIGHT, DAWN, DUSK and CONTINUOUS are recognized.
• Character cues: @John with a leading @ is the most reliable. Standard Fountain all-caps lines (JOHN SMITH, MARY (V.O.))

and one-line forms like Mrs. Robinson: dialogue are read too. On save they are normalized to @Name.

• The all-caps rule is active only inside `INT./EXT.` scenes or when the script contains no Korean at all. In a Korean script,

a lone all-caps line such as OK or TV is treated as a caption or sound effect, not a character.

• If the project language is English, the co-writer, episode review and PDF are in English too (a new project takes the UI language

at creation as its project language).

• If the project language is English, the script PDF uses the US industry-standard format: US Letter paper, Courier 12pt,

standard indents for character names, dialogue and parentheticals, scene headings, names and transitions in caps, and page numbers at the top right from page two (2.). When a speech runs past the page, (MORE) goes at the bottom and (CONT'D) next to the name on the next page; extensions such as (V.O.) and (O.S.) are printed as written. One page is roughly one minute. The cover, planning and episode divider pages use the same paper. PDFs of Korean projects keep the previous format (A4).

• If an all-caps action line is read as a character, start it with ! (!BANG). That line becomes action whatever it looks like,

and the ! is removed. Lines made only of punctuation, such as !!!, are left as they are.

When a line is spoken in another language, put the cursor on that character's cue (or line) and pick a language from Line language in the toolbar. It applies to the whole dialogue block, and video production instructs that line as “spoken in Japanese”. With the cursor on a scene heading, the same spot shows Scene language, which sets the scene's default language. Choose Project language to clear it. The chosen language shows as a small code such as JA next to the cue or scene heading.

When you paste or import a script you can write the same thing as text — [[lang: code]] on the line right after the character cue (or under the scene heading) turns into that setting on save.

INT. CAFE - DAY
[[lang: en]]

@Yuki
[[lang: ja]]
(softly)
おはよう。

• Language codes are short tags such as ko, en, ja, zh, es. If the regional accent matters, add the region — pt-BR

(Brazilian Portuguese), es-MX — and the video direction carries it as well. Without a tag the language is guessed from the script (Hangul → Korean, kana → Japanese, CJK ideographs → Chinese). A single English line in a Korean script is read as English without a tag.

• Languages that share the Latin alphabet — English, Portuguese, French, Spanish — can't be told apart by their letters.

An untagged Portuguese line in a Korean project is directed as English, so set its Dialogue language.

• If a character mostly speaks one language, enter it under Language in the character card's "Video acting & voice settings".

That character's untagged Latin-alphabet lines are then read as that language. Lines written in Hangul or Japanese kana always follow their script, whatever this setting says — for a multilingual character, set the main language there and tag the lines in their other languages.

• A foreign-language line is spoken in that language only if it's written in that language. A line written in Korean with a

parenthetical like (in Portuguese) isn't translated; it's spoken in Korean.

• When you generate a video with dialogue in a language other than Korean or English, you'll be asked to confirm that it hasn't

been verified in production yet. Generation isn't blocked, but check afterwards that the pronunciation and mouth movements match.

• On standard Fountain export the tag becomes an (in Japanese) parenthetical. In the PDF, only English projects (US format)

print the same parenthetical for dialogue in a language other than the project's; PDFs of Korean projects print nothing extra.

AI help — the right panel

The script screen's AI lives in the AI help tab of the right panel (the default tab is Outline). Writing it yourself is the default, and you call AI only when you choose to — even with the tab open, AI never speaks first.

When you're stuck — What's next?

Click [What's next?] at the top of the panel, or type a question and send it (⌘/Ctrl+Enter), and it looks at the text before the cursor, the project canon, and the treatment plan (if any), then suggests 2–3 directions to continue as cards. The suggestions themselves don't write body text (dialogue or action) — you see only directions, and you make the decision. This conversation costs 1 credit per question (the badge next to [Ask] tells you), and the history is kept per episode in this browser (the last 20 turns; delete it with [Clear]). There are also quick questions like "How could this turn?" and "How should this episode end?"

Click [Write a draft from this] on a direction card you like and it writes one scene in that direction (50 credits, the same as the co-writer's scene drafting. If no scene comes back, you get the credits back). The draft arrives in the panel only as a preview; to insert it after the scene where your cursor is, you have to click [Insert after S#n]. Once you click, it appears in the text right away and is saved automatically. The script as it was right before insertion is kept in the version history as "Before inserting draft."

Where episode review and polish results gather

Run episode review from the lower part of the same panel (what the review looks at is in the "Episode review" section below). You can fix the script right away while looking at the notes, and the results stay even if you go to the Outline tab and come back. Results from polishing text you drag-select gather in the same panel too.

On screens where the window is narrow enough that the right panel hides but editing is still possible (roughly tablet width), the far right of the toolbar has "AI help — Episode review" in the ⋯ menu, which opens the review as a dialog, and selection polish results appear as a card at the bottom right of the screen. Mobile is read-only, so this menu isn't there — you can only read the script and leave comments.

When a particular passage bothers you — selection polish

Drag-select text and a bubble appears above the selection (5 credits per call).

• Feedback — 2–4 dramaturgy notes on that passage. Nothing is applied; it's for reading.
• Polish — tidies up only the sentences, keeping the same meaning and format. It doesn't touch your word choices.
• 3 alternatives — three versions with a different feel that serve the same dramatic function.

Polish and alternatives are applied only when you click the replace button in the results (the AI help panel on the right, or a card on narrow screens), and if the original changed in the meantime, it doesn't overwrite it and asks you to select again. Selections spanning multiple lines can break the format, so only feedback is offered for them.

Episode review — what it looks at

The credits for episode review depend on the length of the script, and the exact amount is shown before you run it. Each run covers three things together.

1. Proofreading suggestions — spelling, spacing, and typos, as an "original → fix (reason)" list.

Apply or ignore each one, or apply all. Dialect and colloquial speech in dialogue, character and place names, onomatopoeia, and ellipses are treated as intentional and not corrected. Suggestions that rewrite a phrase rather than fix it are left out. If the same spelling appears more than once in a scene, the item shows "(N places in this scene)", and applying it fixes all of them in that scene only. When a fix can't be applied, the reason is shown — Source changed (edited since the review) · Scene not found (the scene was deleted or its heading changed) · Not saved (it's in the text but the save wasn't confirmed). AI proofreading is a suggestion, not 100% accurate — which is why nothing is applied automatically.

1. Dramaturgy notes — an overall assessment, strengths, suggestions for improvement, and notes per scene. No scores or grades are given.

Click generate revision (5 credits) next to a note and it shows a scene revision reflecting only that note side by side with the original; it's applied only when you click replace this scene.

1. Continuity check — looks for places where the script contradicts the project settings. It's free.

When you want to build the idea together — the co-writer

Use the co-writer on the project home to go from ideation to scene outlines to scene-by-scene drafting, all through conversation. The co-writer isn't opened from the script screen. The advanced guide "AI co-writer and story settings" covers it in detail.

Next: Planning assets — setting up characters, locations, and props. This is a step you must go through before moving on to video production.