Find repeated wording in English drafts with AI
Compare a draft with your own earlier writing and decide which recurring words and sentence habits deserve an edit.
To find repeated wording in English drafts with AI, begin with a collection that actually resembles what you are editing. The voice-baseline skill uses a local Python script to count a watchlist of English words and look for particular sentence shapes. A folder of published columns makes a better reference for a new column than a mixture of short emails and long research reports.
You get counts, not an instruction to remove every occurrence. A deliberate refrain can belong in a piece even if it stands out. Other patterns, such as ending nearly every article with a question, are easier to judge across a collection than on a single page.
What it does
- Build a profile of watchlist words across finished pieces.
- See which words frequently open headings or sentences.
- Compare one new draft with the corpus average per file.
- Identify recurring paragraph endings worth reading aloud.
- Measure again after publishing new work and revising your style notes.
Install and use
- Load the skill in a file-capable agent with Python 3. Put authorised finished texts in one folder and keep the draft and generated report elsewhere.
- Ask, “Measure the writing in my essays folder, then check this new column. Show me which flagged passages to reread.”
- Inspect those passages in context and decide whether a change serves the reader. The script leaves the original prose intact.
FAQ
Source
Made by Huy Tieu. Original skill: voice-baseline, published on GitHub under the MIT licence.
View on GitHub