How the free self-check works
What it measures, what it misses, and why it stays a reading aid.
This page explains the method behind the self-check on the home page: which surface features it looks at, where those features mislead, and how to read the result honestly.
For reading your own writing more sharply. Not for beating a checker. Nobody can promise what a detection service will report, so we don't.
What the list covers
- It looks at surface patterns a reader can also see: repeated sentence shapes, stock transitions, uniform paragraph length, and generic wording.
- It does not identify an author. Nothing here can tell you who wrote a text, or whether a tool was involved.
- Everything runs in your browser. The text you paste is not sent anywhere by this page.
- The list is a heuristic, not a trained model, so it has no probability score to report.
- Short samples are weak evidence. Under roughly 150 words there is rarely enough signal to say anything useful.
- Non-native writing, technical documentation, legal boilerplate, and heavily templated business prose often look uniform for reasons that have nothing to do with how they were drafted.
- No checklist of this kind catches every machine-drafted passage, and none avoids flagging some human writing. Neither outcome proves anything on its own.
- A result is a reading aid you can argue with, not a decision you can outsource.
How to run it
- Read the passage once as a reader, before you look at any checklist.
- Walk the list and mark only the items you can point to in the text.
- Count your marks and read the band on the home page.
- Write down the two or three weakest sentences and say what specifically looks flat about each.
- Fix the writing you can defend changing, and leave the rest alone.
Frequently asked questions
Does the self-check send my text anywhere?
No. It runs in your browser and nothing is uploaded by this page.
Can it tell whether a text was written by a machine?
No. It reports surface patterns. Those patterns are weak evidence about how a text was produced, and they are not proof.
Why does a short sample give a weak result?
Below roughly 150 words there is not enough text for repetition and rhythm to mean much, so the reading stays rough.
Is this a bypass or detector-beating guide?
No. It is a reading aid. It does not tell anyone how to get past a checker, and it is not meant for academic dishonesty.