File
SMF·PROWRITINGAID
Received on
31.07.2026
Reviewed on
28.09.2026
Exhibits annexed
3
Questions
4

SMF·PROWRITINGAID

Can a prompt replace ProWritingAid?

AI writing — grammar, editing and long-form writing

Yes Verdict recorded on 28.09.2026 · Verified on 31.07.2026
Price
$30/moSource: prowritingaid.com · Checked on July 31, 2026
Per year
$360
Build time
One sitting
Category
AI writing
Votes
0 votes
Yes (checked)AlmostNot yet

Exhibit tracking slip

Exhibit A The prompt
Exhibit B What you lose
Exhibit C Why people still pay: polish, integrations, and proprietary language models
Exhibit Q Questions

Verdict

Almost every report ProWritingAid is known for is deterministic text analysis, not machine learning: sentence length variance, adverb density, repeated openers, overused words, readability grades, passive constructions. Those are countable, they need no API key, and a working version runs in one sitting on a local file. The grammar engine is the part that will not match, and the integrations are simply absent — but the reports, which are why people choose this over a plain grammar checker, are genuinely reproducible.

Exhibit B — What you lose

Exhibit A — The prompt

Received on31.07.2026
Build a local manuscript style checker. Deterministic analysis only, no API keys, no network calls.

Input: open a .txt, .md or .docx file. Parse it into paragraphs and sentences with a proper sentence splitter that handles abbreviations, initials and quoted dialogue, because a naive split on full stops makes every report downstream wrong.

Implement these reports, each as an independent module producing findings anchored to character offsets:

- Sentence variety: length of every sentence, plotted as a sparkline per paragraph, flagging runs of four or more sentences within two words of the same length.
- Repeated openers: sentences beginning with the same word within a short window, and the overall distribution of opening words.
- Overused words: content-word frequency against expected frequency, ignoring a stopword list, scoped per chapter and across the whole document.
- Sticky sentences: the ratio of glue words (the 200 most common function words) to content words, flagged above a threshold.
- Adverbs and filter words: -ly adverbs, plus a maintained list of filter verbs (felt, saw, noticed, realised) that distance a reader from a point of view.
- Passive constructions: auxiliary-plus-participle patterns, with the agent identified where one is present.
- Readability: Flesch reading ease, Flesch-Kincaid grade, and average syllables per word, reported per chapter.
- Echoes: any content word repeated within a configurable window of a few hundred characters.
- Dialogue tags: distribution of said versus everything else, and adverbs attached to tags.

Interface: three panes. Report list with a count per report on the left, the manuscript in the middle with findings highlighted, and the selected finding's explanation on the right. Clicking a finding scrolls to it. Every report has a threshold the user can change, and the change re-runs instantly.

Persistence: ignored terms, custom thresholds and a per-project word list stored in SQLite, so a second run does not re-flag things already dismissed.

Out of scope: grammar correction, any AI rewriting, browser or word-processor plugins, cloud sync, and a collaborative editing layer. This tool tells you what is in your prose; it does not write it for you.

Opening prefills the prompt — press enter to run it.

Exhibit B — What you lose

  • B.1 the browser, Word and Google Docs integrations
  • B.2 grammar accuracy tuned against years of accepted and rejected suggestions
  • B.3 syncing a document and its ignored terms between machines
  • B.4 the fiction-specific reports on dialogue tags and pacing by scene

Prior art

Exhibit C — Why people still pay: polish, integrations, and proprietary language models

Because it is already inside the editor where the writing happens. A separate application you have to remember to open is a report you run once and then forget about.

Questions

Can I bring my ProWritingAid settings across?

No. Ignored words and custom style rules live in your ProWritingAid account and are not exportable, so thresholds and word lists get rebuilt here. It takes about an hour of tuning against a chapter you already know well.

Does it check grammar too?

No, deliberately. Grammar correction needs either a large rules engine or a model, and doing it badly is worse than not doing it. If you want that layer, the catalogue's LanguageTool entry covers a self-hosted option that pairs well with this one.

What does it cost to run?

Nothing. It runs locally with no API key and no network access, which also means a manuscript never leaves the machine — the one thing this build does strictly better than the subscription.

What is the one thing that does not survive the rebuild?

Being where you write. ProWritingAid checks as you type inside Word, Docs and the browser. Here you finish a draft, open a separate application, and run reports on it, which turns continuous feedback into an occasional audit.

Receipt

Already built this yourself?