Release Notes
Version: "0.38.0"
- A project holds many scripts. Library, Scripts lists every script in a project, and any of them can be a main script, in sequence, for episodes and sequels. Opening a .fountain, .fdx or PDF with a project open adds a script to it instead of replacing anything, and File, Duplicate Project clones the whole project. PDF import asks which pages to read and whether the file is a screenplay or a document, and works on iPhone and iPad, scanned pages included.
- Scenes, documents and a breakdown in the Library. Every scene is a page with its synopsis, shots, Breakdown and the notes tagged in it; scripts expand to their scenes, and scenes move, copy and paste between scripts. A Documents folder takes PDFs, Word and text files and shared Google Docs links so the assistant can read them, plus manuscripts for prose. The project report prints or exports what every scene needs, by department. Pictures, video and audio get thumbnails, paste and drop, and one copy per device.
- Storyboards grow up. A storyboard is a folder of shots, nested as deep as you like, timed against its own script and linked to the scenes and outline beats it covers. Generate works across script, outline and storyboard with a preview first, Storyboard this starts from a scene, a contact sheet adds frames across a scene, Scenes from outline gives every beat a scene, synopses write back to the script, and a board prints and exports as a PDF.
- 3D blocking. Stage a scene in 3D from the Scenes tab: sets, props, a mannequin with built-in moves, 3D models from the Library (.glb, OBJ, FBX, USD and more), textures, skies, colour looks, sound cues and dialogue captions on a timeline. Stage with AI builds a first pass from the script and Refine with AI looks through the cameras and fixes it. Panels take their frames from the blocking and warn when stale. A location keeps a set that its scenes share, there is a live Overhead view, and on the desktop Open in Blender and Set up for Blender hand the scene over.
- Table reads and a teleprompter. View, Teleprompter shows the script in large type with the line being read mid-screen. Mark the parts people in the room will read and the read waits for them. AI voices work in read aloud, table reads and the teleprompter.
- The assistant. Synopsis lines are story structure: a = line summarises what follows it, Edit, Write synopses fills in the missing ones, and the outline lines up with the script without a model call. Every prompt is sized to the model and device. Selected-text tools no longer overwrite the screenplay. Over MCP an assistant can read story modes, the help manual and the 3D blocking, with the frame as a picture, and asks you once before its first 3D edit. On-device OCR and voices are built in. Chat scrolls smoothly on long threads.
- Find. A Match Case toggle in the find bar and in Edit, Find, off by default, so a phone's auto-capital no longer hides matches. The find bar clears the logo on iPhone. Desktop menu toggles show their tick.
- Settings, Beta. Bone Yard, Section headers and the Tab button are opt-in features there, and a /* boneyard */ in a Fountain file reads back as a Bone Yard instead of being dropped. Pro members can turn on the Sequencer: a timeline of your shots with sound, camera moves, video takes and an MP4 render.
- Support and billing. Replies to a support ticket reach you in the app and by email, with a conversation screen to answer from. A subscription can be billed by one store at a time, and switching is handled on the server.
- Translated into 28 languages, with the keys that were English-only filled in.
- Page breaks by the rules. A speech or action block that runs over a page now breaks only at the end of a sentence, with at least two lines on each side, and a speech that breaks carries (MORE) at the foot of the page and NAME (CONT'D) at the top of the next, extensions like (V.O.) included. That holds when the break falls between a parenthetical and its line, too. The PDF and the Print Editor now paginate the same script onto the same pages, a character cue is never left alone at the foot of a page, and the Print Editor no longer overlaps text where an element splits.
- Deleting stays calm. Holding Backspace with Autocomplete while typing on no longer sends a request between every repeat, and no longer ends in the app restarting after an error on iPad. The autocomplete and a ✨ suggestion wait until you stop deleting, then pick up from where the caret is.
- One AI at a time. The autocomplete ghost steps aside while a ✨ suggestion or a command is working or has an answer on screen. A proposed change that comes back by undo, by reopening the file or from a collaborator shows its review bar again, so it can always be accepted or rejected.
- Your own assistant, at the controls. On the desktop, Settings, AI has a new switch, Allow desktop assistants to control EpicDraft, off by default. Turn it on and an assistant you already pay for, such as Claude Code, Claude Desktop or Gemini CLI, can read the open screenplay, keep its Library and run coverage. Anything that overwrites work is approved by you in EpicDraft first, the approval and the Undo entry both name the assistant that asked, and a chip in the status bar shows who is connected and can cut them off. The settings card shows the one line to run to connect a terminal on this computer.
- The assistant knows where you are, and remembers. Ask Where am I and it answers from the caret: the scene, the outline beat that scene plays, and the matching place in the treatment. What's connected here follows the Library's links out from the page, scene or character under the caret. Remember this saves a note to the project, kept per chat session and per model, What do you remember brings the notes back, and every chat turn opens with what it remembers. Search the story searches by meaning for plot points, a character's voice or behaviour, the story's progression and what the coverage said. A Memory section in the Library lists the notes by provider, with editing and forgetting, and Settings, AI, Share AI memory with collaborators decides whether they reach the assistants of the people you share a script with.
- Images that survive the device. Pictures and media in the Library display on the desktop again, a thumbnail that cannot load says so instead of showing an empty tile, and the media viewer has a Download button. Media is now stored in the project itself, so an image added on one device appears on the others and comes back after the device's local store is cleared. With Cloud Sync on, the bytes travel with the project.
- Tab only ever accepts. A suggestion starts from the ✨ button or Edit, AI Suggestion, never from a key, so an iPad's auto-capitalised Tab no longer starts one you did not ask for. Tab lands exactly what the ghost shows at that instant, mid-sentence and mid-stream, and the model keeps going; with nothing on offer it indents as usual. Editing the line a suggestion is streaming into, by undo, an Edit menu cut, Scribble or the on-screen keyboard, cancels it instead of letting the ghost text write into wherever the caret went, and deleting a page down to only headings and cues no longer throws you back to the file list.
- Suggestions write names the way the script does. An established character is written in action and dialogue as the page spells them, Mara rather than MARA, and is not introduced again. Changing an element's type no longer re-asks the model. Next action and Next dialogue write one element, not a scene.
- Commands that fit their model. On Ollama the app runs with a 16k window it actually budgets against, and a command that cannot fit is refused before it runs rather than losing the head of its brief in silence; a large highlight is sent once; a continuation the model refuses is re-fitted and retried. A document-wide answer that stops short can be read and copied instead of vanishing. Japanese, Chinese, Korean and Thai scripts are measured at their own rate, so a vertical script fits the same window an English one does.
- Chat. Chat with a model on your own device works offline and with an expired session. A turn refused for size keeps the tool steps that already ran, so a wiki page is no longer appended twice on the retry. On an on-device model the transcript and the project digest are sized together, so there is room for both. Gemini chat works.
- Editor. An empty caret line under the suggestion chip is one line tall again, so the selected-line highlight and the floating toolbar sit where they should.
- Suggestions that write screenplay. A Shift+Tab suggestion comes back as script, not prose: dialogue is what the character says, an action line stays an action line, and a character the script has already met is not introduced again with an age. Change the element's type while a suggestion is on the line and it is asked again for the new type.
- Tab lands what has arrived. While a suggestion is still streaming, Tab drops in every complete paragraph so far and the model keeps going; once it has finished, Tab lands the rest. Escape is the only cancel. On an empty line the suggestion chip and the autocomplete ghost now sit on the caret's own line in every browser and on iPad, instead of the line before or after.
- The assistant knows where it is. A command is briefed with one scene number wherever a scene is named, sees the whole scene it is continuing rather than its last few lines, and fits its brief to the model's window. Finish the script writes one scene per turn instead of failing for size, and chat sends the model a transcript it can hold.
- Team rooms for everyone in them. Members of a shared script now see each other's names, "/" commands as soon as @ai is addressed, a working bubble while a teammate's @ai prompt runs, and a byline naming who invoked the reply and on which model. The Send button no longer sits on the addressing chips, and the assistant no longer reads its own replies back as yours.
- Settings follow the account. Signed in, Cloud Sync, Sounds, Paper and the other app preferences follow the account: change one on your phone and it changes on your desktop while the desktop is running. Zoom stays with the device.
- On-device dictation on the desktop. The mic button works on Mac, Windows and Linux with a Whisper model (Tiny, Base or Small) that downloads once and runs entirely on your machine. Downloaded speech, image and voice models can be switched off or deleted in Settings, AI.
- iPad and iPhone. Coming back to a script after the app has sat idle no longer leaves the editor on a half-full progress bar, and the app lock survives a reload. Deleting a script from Settings or the Library returns to the file list instead of a Something went wrong screen. On-device models are budgeted from the memory the system really leaves the app, so a model that fits is no longer refused, and a command the device cannot hold is refused before it runs rather than coming back cut off.
- Desktop. Closing the main window hides it to the tray; Quit in the tray menu or Cmd+Q still quits. The side nav opens collapsed to its icon rail on web and desktop and its file rows line up. Zooming in past 100% centres the page and scrolls to its true right edge.
- Editor fixes. Switching files no longer strands the editor on its loading skeleton with Thinking... in the status bar, and opening the file list on the no-script screen no longer blanks the view.
- A new sample catalogue. Seventeen original sample scripts by Avery Quinn, every one written for a general audience with its own cast. The Midnight Line is the script a first run opens and the one the Narrative card points to, and Nine Tenths of the Law (comedy) and Salt and Static (science fiction) join the bundle. House Rules and Helix are retired.
- Parentheticals keep their case. A parenthetical such as (to Des) was saved in lowercase and never re-cased, so the proper noun was gone after one save. The file now keeps the text exactly as typed.
- Dialogue that opens on an ellipsis. A line such as "...That was good." right under a parenthetical or a cue is dialogue again, not a forced shot, on first parse and after a Fountain round trip.
- No more stuck loading screen. Opening a script could leave the editor on its loading skeleton, with the outline filled in but no text, until the app was restarted. The editor now always comes up.
- Usage reporting now sends what it promised. App opens, session time, saves, opens, exports, update checks and errors are reported as anonymous events under the existing telemetry setting, with the same opt-out in Settings, Application, Privacy.
- The assistant follows your outline. Next scene, Continue from here and Expand beat now know which beat the last scene played and which comes next, so a continuation picks up the story where the outline says it should. A continuation started mid-script is placed inside that scene, before the one already written.
- Print shows script, not scaffolding. Synopses join inline notes, the boneyard and section headers in staying out of the Print Editor and the PDF, and the controls that create them grey out while Print is the view. Hidden blocks no longer reserve a blank line on the page, a forced page break ends its page cleanly instead of leaving a stub sheet, and a typed (CONT'D) cue is visible in the editor again.
- The reader, in three steps. Read the script full width with one scrollbar, review your notes and choose which ones the author sees, then write the coverage. Nothing is sent until you send it. Authors set a content rating in Sharing, and R, NC-17 and unrated scripts ask a reader to confirm they are 18 or older before the first page. The reader's lists show each script's rating.
- Cue extensions autocomplete. Type ( after a character name for (CONT'D), (V.O.), (O.S.) and the rest, in the document's own language, so a French script offers (SUITE). Lowercase and smart quotes complete too, and accepting onto a cue that already speaks no longer wedges an empty line under it.
- Floating toolbars fold away. Both caret-following toolbars fold into a drawer behind their toggle and park in the margin, so a folded bar never sits on the text you are writing. The choice is remembered per device.
- App preferences belong to the device. Cloud Sync, Sounds, Paper and the side-nav settings now follow the device rather than the account, a change in one browser tab reaches the others, and each setting says so.
- End-of-script buttons are opt-in. The continuation buttons under the last page are off unless you turn them on in Settings, AI, Defaults. Continue from here stays in the AI command dialogue. The "no ending yet" note is gone.
- Mac updates install again. Restart and update now works on the Mac download. Your files and settings move out of the old container once, on first launch, and an update that fails says why instead of doing nothing.
- Smoother scrolling. Long scripts no longer tear or jump when scrolling up, and the scrollbar reports the true length.
- Tablets and phones. The AI command dialogue and its review bar sit above the on-screen keyboard on iPad. On-device models are sized to the memory they really get, so an AI command sends a fraction of the context it used to. Print view is available on phones, the phone file list opens with the desktop onboarding, turning dual dialogue off no longer selects the whole script, dual dialogue can be toggled from a highlighted selection, and cut, copy and paste no longer carry the autocomplete ghost along.
- Editor fixes. Moving blocks keeps the blank line at the seam and works from inside a dual dialogue; the dual dialogue toggle stops guessing where the right column went; inline notes end with a blank line; a real Clear Color button in the colour and highlight pickers; whitespace the page never prints is hidden; the AI suggestion stays on the caret's own line and clears when the caret is gone; lyrics keep their leading space; the file header keeps its key order; the Library panel lines up on one grid; the macOS Control glyph is the real one; Sharing and Submissions are back in the desktop File group; a cascade of errors now recovers with a refresh or a file reset.
- Documentary template in single-column A/V.
- Fixed an intermittent crash on launch on iPhone, iPad and Mac. When a resource the app was still loading got cancelled during startup, the app could quit a moment after opening. It now carries on.
- Vertical writing. Japanese and Traditional Chinese screenplays can be written top to bottom, right to left, the way a 脚本 is printed: cue and speech flow as one column with 「」 supplied for you, ○ scene headings are recognised as typed, page breaks read （つづく）, and the clock counts characters instead of pages. Native vertical PDF export, and a Writing Direction setting on desktop and mobile.
- A linter you can work through. Craft checks now know each language's conventions, so a Spanish CORTE A: or a German (FORTS.) is not a warning, and seventeen new Hollywood spec-format rules cover parentheticals, (CONT'D)s, near-duplicate names and locations, character introductions, transitions, scene length and the title page. The Review tab and the status bar walk you from issue to issue in the editor, any rule can be silenced for one script, and the assistant can run the same checks and fix what you approve.
- The outline, checked against the draft. Ask the assistant for Outline progress and it maps every scene to the beat it dramatises, shows where the draft drifts from the plan or grows past it, and names the next story point to write.
- Synopsis lines brief the assistant. A synopsis line (one that starts with =) above a scene heading is that scene's summary, and above a passage it briefs what follows. The assistant reads them wherever it reads the script.
- Storyboards linked to the script. Click a range of script lines to attach it to a panel, follow a panel's link back to the script, and start a storyboard from a scene heading's menu.
- Cast lists for television and stageplay. A cast list element directly under the scene heading, the multi-camera convention, offered by the television and stageplay formats.
- A/V columns that hold their shape. One blank line inside a column is spacing, two end the block, the PDF wraps each column at its own width, adjacent blocks merge, and Format > Columns has Alt+Cmd+1, 2 and 3 shortcuts.
- Plain Fountain export. Save a Copy as a portable .fountain file.
- Game scripts. Logic tags in the gutter, an Add choice action, intended endings, a Map that fits the panel with zoom, a Play view that lists the choices a condition has closed, a Vars tab, and a Game screen on phones.
- Autocomplete while typing, experimental. Turn on Settings > AI > Autocomplete while typing and a fast model offers the next few words as ghost text at each word boundary. Tab accepts, typing past it lets it go, and it works on wiki pages too. Off by default.
- Your free AI trial, out in the open. New accounts see the 14 free days promised at signup, counted down while they run, and an upgrade offered when they end. Pro granted through an invite link can now carry a term.
- Cut and paste that keeps its shape. Pasting several lines keeps every element's type and the blank lines between them, wherever the caret was, on phones as well as desktop. Copy and cut no longer lose the clipboard when the editor is mid-update.
- Phones and tablets. Autocorrect works again in scene headings and character names on iOS, the per-file Language settings, including writing direction, are on mobile, and the hardware Back button closes the drawer and steps back one screen at a time.
- Faster exports. PDF export of Japanese, Chinese and Korean scripts takes seconds instead of most of a minute.
- Korean and Chinese screenplay vocabularies, first drafts, so scene headings and transitions in those languages are recognised as typed.
- Your telemetry choices stick. A consent answer survives an iOS storage cleanup, and collected events are pruned on the schedule the privacy policy promises.
- Fixes throughout: an all-caps shot line before a blank line is action rather than a character cue, a synopsis no longer gains a space on every save, the scene-heading menu stays on screen in vertical writing, opening a wiki page from the editor on a phone no longer leaves the editor behind it, and admin access follows the account role rather than a fixed list.
- Bring your own AI. Keep the assistant on local models, the default, or connect your own OpenAI, Anthropic, Google or xAI account with your own key. Pick a different model per task, so chat, rewriting and coverage need not share one.
- Images and speech. Generate storyboard frames and concept art from the wiki page you are on, and have a scene read aloud. Both are opt-in and off until you turn them on.
- A global Stop. One button halts any running AI feature, wherever it was started from.
- See it before it lands. Editor commands arrive as a staged diff you accept or reject, the editor shows exactly where a command is aimed, and the inline suggestion is back — on request only, never on a timer.
- File history. Every project keeps a version history with automatic checkpoints, one before each AI edit, with its own settings per file and app-wide. History stays on your devices and is never uploaded.
- New formats. Stageplay, ads and branded content, documentary two- and three-column A/V layouts, and game scripts with includes and cross-script branching.
- The project graph. A weighted map of how scenes, characters and pages connect, in the Library — and the assistant can use it to answer from the part of the project you are actually in.
- Wiki pages grew up. Print any page, place images inside table cells, generate art from the page's own context, and answer the six key-character questions on protagonist and antagonist pages.
- Writing, faster. An IDE-style ghost preview for autocomplete that Tab accepts, hints showing which shortcut sets each element type, and every document remembers where you left off.
- Connect your own tools. EpicDraft speaks MCP, so the assistant can reach the tools you already use, and other apps can reach your project.
- A free tier that stands on its own, with Pro covering file history, the project graph and image generation. Pro Lifetime is a one-time purchase: the app and every offline Pro feature for good, with five years of Cloud Service included.
- Fixes throughout: the Return key and autocorrect on phones and tablets, printing, sync, dictation, the editor toolbars, and accessibility work across the app.
- The Library. The separate story planner is gone, folded into your project's wiki alongside characters, locations and outlines. Outline templates moved with it, and the whole thing now sits under Library.
- Character sheets and lookbooks. Character pages grew into proper sheets, and locations and story-bible pages gained the same lookbook and reference panels.
- Storyboards that follow the script. Panels anchor to scenes, with timing and automatic naming.
- AI that sees the whole project. One shared context, so the script, the planner and the wiki know about each other. Suggestions now arrive in the document's own format and sew into the page instead of landing as a block of text, and the assistant understands what you have selected.
- Decide how adventurous the AI is. A Creativity setting that escalates if you ask again, guidance on which local model suits which task, and the option to share one model between your own devices.
- A real chat composer. Rich text, a searchable command menu, and a composer that grows and locks while the assistant works.
- Tables and lists in the editor. Proper editing for both, and chat markdown that renders the way it was written.
- Links, faster. A link button on the main toolbar with Cmd+K, and opening a link is now separate from editing it.
- Scenes on mobile. A Scenes screen in the drawer that jumps the editor straight to any scene.
- One Review tab. Reports, Submissions and Sharing merged into a single place, with file security settings on the Title tab.
- Watermarks when printing. Add custom text to a printed or exported PDF.
- Telemetry you control. Consent settings and a first-run notice, with nothing collected unless you turn it on.
- A 14 day AI trial for new accounts.
- Accessibility work across the editor toolbars, per-format file icons, CSS named colours in the pickers, an expanded Format menu with more hotkeys, and a long list of fixes across AI, sync, mobile keyboards and printing.
- Story Wiki — every project now has a wiki beside the script: characters, locations, outlines, notes, and storyboards, organized into folders with breadcrumbs and page tabs. Type
[[ anywhere to link a page, and follow links straight from the screenplay.
- Storyboards — lay a sequence out as panels, each with its own image, shot, and action notes.
- Reference media — drop images, video, and audio onto a wiki page. Media stays on your own devices and syncs directly between them; we never host it.
- Story modes — start a project as Narrative, Television, Documentary, or Game and get page templates, structure, and AI help suited to that format. New sample scripts and format examples are there to explore from the first-run screen.
- On-device AI — run a model entirely on your own computer or phone: no account, no cloud, nothing leaving the device. You can also point the app at a model server on your own network, or keep using a hosted one.
- A model per task — pick a different model for chat, rewriting, notes, or coverage, and switch between them without leaving the editor.
- AI that knows your project — the assistant can read your wiki, characters, and outline for context, and write new pages back, clearly marked as AI-written.
- AI tasks — kick off a coverage report, a consistency check, or a character interview; it reads the whole project in the background and files the result as a wiki page, with progress and cancel.
- Connect your own AI tools — attach external tool servers to the assistant from Settings → AI.
- AI usage log — token counts and estimated cost for every request, kept on your device.
- Reading Exchange — trade coverage with other writers: read someone else's script to earn a read of your own, with an agreement gate before anyone opens your pages.
- Redesigned desktop layout — the file area folds into a two-column side navigation that you can put on the left or the right.
- Save As on every platform — choose the format and where the file goes, including on Android.
- Search the manual — full-text search across the in-app help.
- A branded launch screen instead of a blank window on cold start, plus dozens of fixes across mobile keyboards, Android printing and navigation, purchases, and sync.
- Reimagined AI — a rebuilt AI engine with faster responses and a screenplay-aware assistant that understands scenes, characters, and context as you write.
- Real-time collaboration — invite teammates with viewer, reader, and editor roles, transfer ownership, and protect sensitive drafts with password encryption.
- Team chat — every shared script now has a real-time chat with @mentions, so you can talk through notes right next to the writing.
- Notifications inbox — keep up with mentions, coverage, reading activity, and read-access requests, with per-event and per-script controls.
- Script Reader & Coverage — share a read-only link, collect inline notes page-by-page, and track feedback from a coverage dashboard with leaderboards.
- Write in your language — the interface is now available in English, German, Spanish, and French, with more on the way. You can also give each script its own language — including right-to-left Arabic and Hebrew and East Asian scripts — so scene headings, transitions, and continuation text follow local conventions.
- Biometric unlock & app lock — open password-protected scripts with Face ID, Touch ID, or Windows Hello, and lock the whole app automatically after you've stepped away.
- Dictation — write with your voice on mobile; spoken text drops straight into the editor.
- Public author profiles — claim a handle and share your work at a clean
/author/you URL.
- Per-file formatting — set paper size and element spacing on a document, with a new floating element toolbar (toggle with Cmd/Ctrl+J) and spacing guide lines.
- Smarter imports — AI-enhanced PDF import that reads scanned scripts.
- Dozens of stability, security, and mobile UI fixes.
- In-app purchases on mobile — subscribe directly on your device, with restore-purchases and subscription management built in.
- Refreshed mobile interface.
- Added subscription policy links and a cleaner pricing chooser.
- Hardened billing verification and receipt handling behind the scenes.
- Added an in-app feedback and support channel so you can reach us without leaving the app.
- Updated for the latest Android platform.
- Polished app store listings with new screenshots.
- Automatic page breaks — dialogue and action that crosses a page edge now splits cleanly, matching the printed page exactly.
- Folder Sync — keep your scripts mirrored to a local folder on your computer automatically.
- Cloud Sync settings — control how and when your work syncs to the cloud.
- Subscriptions now run through a smoother checkout.
- New help center and blog, with dark mode and search.
- Spell check, auto-correct, and auto-capitalize in the source editor, plus inline image and link support.
- More reliable device-to-device sync on mobile.
- Peer-to-peer sync — collaborate device to device, without your script passing through the cloud.
- Smoother, more responsive collaboration cursors.
- Introduced new pricing and enabled payments.
- Fixed the grey-screen flash and made opening a file into the editor feel seamless.
- Cloud sync with live cursors — see collaborators' cursors move in real time as you write together.
- New loading splash screens and a smoother startup experience.
- Added a Medium theme and a choice of UI style.
- Open plain-text
.txt scripts, and share files directly from mobile.
- Automatic reload after idle to keep your work current, plus a batch of print and iOS fixes.
- New document engine — rebuilt from the ground up for faster, more reliable saving and syncing.
- Sign in with Apple on mobile.
- In-app notices to keep you posted on updates and status.
- Better PDF import, export, and printing, with improved iOS support.
- Added Story Atlas notes and cleaned up genre metadata (primary and secondary genre).
- Story Outliner — plan your script with draggable beats and structure templates, now working on mobile too.
- AI upgrades: scene generation, plus Style & Tone controls to shape the voice of your writing.
- Live word counts as you write.
- New theme support across the app.
- macOS and iOS builds are now code-signed. Fixed a bug that reverted the title while editing.
- Added the account menu with sign-in, sign-out, and subscription controls.
- Scanned-script reading now works fully offline.
- Improved Linux packaging across .deb, AppImage, and .rpm builds.
- Various download and cross-platform build fixes.
- The first public beta of EpicDraft — write, format, and export screenplays with Fountain across desktop, mobile, and web.