Paste a post, get a score from 0 to 100 and the exact lines a reader would recognize as machine written.
58
AI phrasing score
Example
AI phrasing shows
Several model formulas in a row. Readers start to feel it.
You have used your 3 free scans for today.
A free MagicPost account raises it to 50 a day, and runs the same scan on your drafts while you write.
Create a free MagicPost accountGrading a post needs JavaScript. The example below was graded with the same scan.
Rewrite it with MagicPostAn example post, graded with the scan you are about to run on yours.
The post we read
What gives it away5 lines flagged
It's not about volume. It's about resonance.A negation that re-labels the same idea instead of adding one.
Here's what nobody tells you about building an audience here:A reveal opener that would fit any topic on the feed.
Same team. Same product. Different approach.A run of very short sentences a person would have joined with a comma.
The key is to stop writing for the feed and start writing for one person you actually know.A frame like “the key is” or “stop X, start Y”, true of any job.
That's not a growth hack. That's a discipline.A negation that re-labels the same idea instead of adding one.
Nothing flagged. No model formula in this post.
Language fluency96100 means natural and idiomatic.
Read as English
3 free scans a day, no signup
The taxonomy
Every family the grader looks for, and the phrasing that triggers it. The ones found in the post above are marked.
The MagicPost LinkedIn post grader rates a post from 0 (no visible AI phrasing) to 100 (dense AI phrasing), and quotes the lines behind the rating, such as “It's not X, it's Y” or “Here's what nobody tells you”.
| Tell | What it looks like | Why it stands out |
|---|---|---|
| It's not X, it's Y | “It's not about volume. It's about resonance.” | Drop the negative half and the point survives untouched. That is the test, and it is the most repeated reflex in model writing. |
| “Here's what nobody tells you” | “Here's what nobody tells you about X” | A promise of a secret, followed by advice anyone could have given. Swap the subject and the sentence still works. |
| Advice that fits anyone | “The key is to stop X and start Y” | It would be true for any job, any product, any week. It carries nothing that only you could know. |
| Chopped sentences | “Same team. Same product. Different approach.” | Three or more very short sentences where one would read better. The chopping is the tell, even when each one adds a fact. |
| Model vocabulary | “delve”, “in today's fast paced world” | Vocabulary models reach for and people almost never type. A single word can date a post. |
Catch these while you write, not after you publish. MagicPost scores every draft in the editor.
Write it clean with MagicPostHow the grader works
A pattern scan runs first and hands its findings to a language model, which judges the phrasing with that evidence in front of it. Both passes run at the same time, so one grade is one wait.
Every tell points at the exact words in your post. Nothing is reported without the line that triggered it, so you stay the judge of whether the call is fair.
Closing questions and P.S. calls to action earn engagement on LinkedIn, so the grader stays quiet about them. Same for the habits our impressions data found reach neutral.
Your next step
The grader tells you what to fix once the post is written. MagicPost writes from your own published posts, so the formulas above never reach the draft.
A voice the scan does not flag. MagicPost learns how you already write and drafts in that register, not in the one this page is built to catch.
The same scan, inside the editor. Your draft gets the AI phrasing score as you write, with every reported line highlighted where it sits.
50 scans a day, free. An account lifts the daily limit and keeps each scan attached to the post it belongs to.
FAQ
Phrasing, not content quality. A score near 0 means nothing in the wording looks machine made. Above 60 means several model formulas repeat across the post. A weak post can score 0, and a genuinely good post can score high if it leans on those formulas.
It reports how a post reads, not who wrote it. No tool can prove that a model produced a text, and we do not claim to. What you get is the list of formulas a reader would notice, quoted from your own post.
Because closing questions and P.S. calls to action work on LinkedIn. The grader stays quiet about the habits that earn engagement, and about the ones our impressions data found reach neutral. It only reports what makes a post read like a template.
Three a day from this page, with no account. A free MagicPost account raises it to fifty a day and runs the same scan on your drafts inside the editor, line by line. Write LinkedIn posts in your own voice with MagicPost
Join 100k+ users
Get started
MagicPost drafts from the posts you have already published and grades every draft as you write, so the formulas above stay out of your feed.
Try MagicPost
