Paste any text (a news article, a politician's statement, a pastor's sermon, a corporate email) and the engine detects manipulation patterns in the rhetoric. It marks every matching phrase where it sits, names the shape the patterns add up to, and shows who tends to gain from that shape. It does not determine truth.
Runs entirely in your browser. Nothing is sent to a server.
Specimen intakeclient-side · no upload
Load an exampleComposites written for this page. Not quotations.
0 chars0 words
Ctrl + Enter
Deceit · The Scanner · Field instrument
Pattern analysis report
Generated at deceit.media/scan. The analysis ran in the reader's browser; the text was never uploaded. Re-run it any time by pasting the specimen below into the same page.
▌ Deceit · Pattern Analysis
Open
Narrative type
Benefit direction:
0/ 100
Signal index, not a truth score
How to read this report
Narrative type
The overall shape the detected patterns add up to. The signal profile shows how close the runners-up came.
Confidence
An index from 0 to 100 of how much matching language the rules found. It is not the probability that the text is false. A true claim inside a manipulative frame can score high, and a false claim in plain language can score low.
Benefit direction
Who tends to gain when this shape is used, inferred from which patterns fired.
Numbered marks
Each mark in the specimen is a phrase that triggered a rule. Its number points to the detected pattern with the same number. Two numbers on one mark mean two rules read the same words.
Not this when
Every pattern carries the Register's note on when the same words are innocent. Read it before treating a match as a finding.
Method
22 detection rules from the Deceit Register, each keyed to a named pattern with known verbal tells, ran against the text inside the reader's browser. The scanner names the shape of the rhetoric. It does not decide what is true.
Signal profile
Narrative shape
Benefit direction
Where the signals fall
OpeningClose
Detected patterns
Ranked by confidence. The number on each card is the mark it leaves in the specimen below; press it to see only that pattern's marks.
No patterns detected. The text may be straightforward, or its signals sit below the detection threshold.
Specimen, annotated
Showing marks for only.
Flagged phrases
Every phrase that fired a rule, most repeated first. Select one to find it in the specimen.
Who benefits
Counter-questions
Questions that expose the framing. Each one targets the detected pattern named beside it.
Missing context
How the scanner works
The engine holds 22 detection rules, each keyed to a named entry in the Deceit Register. A rule is a set of verbal tells: exact phrases, grammatical shapes such as a passive sentence with no actor, and framing moves such as a deadline asserted without evidence.
When a rule fires, the engine marks the phrase in the specimen, scores how much matching language it found, weighs which narrative shape the firing rules add up to, and infers who tends to gain when that shape is used. Every finding carries the Register's own note on when the same words are innocent.
The analysis runs entirely in your browser using client-side JavaScript. Your text is never sent to a server. The scanner is the same engine that powers the editorial desk, made available as a public tool.
The scanner detects patterns, not truth. A true claim inside a manipulative frame is still manipulation. A false claim without pattern signals is not what this tool catches. Use the scanner for what it is: a lens for the shape of rhetoric, not a verdict on the facts.