> ## Documentation Index
> Fetch the complete documentation index at: https://docs.inkwell.app/llms.txt
> Use this file to discover all available pages before exploring further.

# v1.4.0 — PDF import, rebuilt from the ground up

> A brand-new PDF import engine that understands your script's layout — dual dialogue, stage plays, and shooting scripts now come in clean — plus writing streaks and condensed sides.

<Note>
  Released **July 23, 2026** · [Download the latest version](https://inkwell.app/download)
</Note>

## Highlights

PDF import got a complete rebuild. Instead of reading a PDF line by line and
guessing, Inkwell now looks at **where text sits on the page** — the way you
do — so scripts that used to come in scrambled now import the way they were
written. Dual dialogue stays side-by-side, stage-play formats keep their
dialogue, and shooting scripts keep their scene numbers. Imports are faster,
too.

And Inkwell now keeps your **writing streak**: see your run of writing days
at a glance, and tell Inkwell which days should count.

## New

* **Writing streaks.** Your writing-day streak now shows with your writing
  stats, with milestones as you hit them. In **Settings ▸ Writing** you can
  set your cadence — which days of the week you intend to write — and
  forgiveness, so a planned day off doesn't break your run.
* **Condensed sides (cue scripts).** When exporting sides, pick performers
  from a multi-select dropdown and export a condensed cue script — with
  optional yellow highlighting and a margin bar on your performers' lines,
  just like marked-up table-read sides.

## Improved

* **Dual dialogue imports correctly.** Side-by-side speeches used to import
  interleaved or collapse into one speaker. They now come in as two clean,
  paired speeches — every time.
* **Stage plays and theatrical formats.** Scripts with full-width dialogue
  and centered character names (common in stage-play templates) used to
  import as a wall of action. Inkwell now learns each document's layout, so
  dialogue is dialogue.
* **Reading order is trusted again.** Some PDFs put a character's name in
  the middle of the sentence beneath it, or split one line into scrambled
  fragments. Imports now follow the page's real layout, so sentences read
  in order.
* **Shooting scripts keep their scene numbers**, `SCENE 1`-style theatrical
  slugs are recognized as scene headings, and `OMIT` slugs come in clean.
* **Page-break speeches are stitched.** A speech split across a page break
  with `(MORE)` / `(CONT'D)` now imports as one continuous speech, and
  stray `(CONT'D)` markers no longer pile up on character names.
* **Cleaner text.** Running headers, doubled page numbers, and revision
  stars no longer leak into your script's text, and hyphenated compounds
  like "screw-ups" survive line wraps intact.
* **Faster imports.** The new engine imports large scripts noticeably
  faster than before.

## Fixed

* Transitions like "FADE TO BLACK" without a trailing colon are no longer
  mistaken for character names.
* Shouted all-caps dialogue ("TEN SECONDS.") no longer imports as a
  character cue.
* Your local file paths are no longer stored with writing-session data —
  sessions are tracked with an anonymous identifier instead.

***

<Tip>
  See [all releases](/whats-new/overview) for the full history.
</Tip>