| 99.15% | Adverbs in dialogue tags | Target: ≤10% dialogue tags with adverbs | | totalTags | 23 | | adverbTagCount | 3 | | adverbTags | | 0 | "She turned away [away]" | | 1 | "Eva said carefully [carefully]" | | 2 | "Eva said finally [finally]" |
| | dialogueSentences | 59 | | tagDensity | 0.39 | | leniency | 0.78 | | rawRatio | 0.13 | | effectiveRatio | 0.102 | |
| 90.84% | AI-ism adverb frequency | Target: <2% AI-ism adverbs (58 tracked) | | wordCount | 1638 | | totalAiIsmAdverbs | 3 | | found | | | highlights | | |
| 80.00% | AI-ism character names | Target: 0 AI-default names (17 tracked, −20% each) | | codexExemptions | (empty) | | found | | |
| 100.00% | AI-ism location names | Target: 0 AI-default location names (33 tracked, −20% each) | | codexExemptions | (empty) | | found | (empty) | |
| 90.84% | AI-ism word frequency | Target: <2% AI-ism words (290 tracked) | | wordCount | 1638 | | totalAiIsms | 3 | | found | | | highlights | | 0 | "footsteps" | | 1 | "silence" | | 2 | "familiar" |
| |
| 100.00% | Cliché density | Target: ≤1 cliche(s) per 800-word window | | totalCliches | 0 | | maxInWindow | 0 | | found | (empty) | | highlights | (empty) | |
| 100.00% | Emotion telling (show vs. tell) | Target: ≤3% sentences with emotion telling | | emotionTells | 0 | | narrationSentences | 113 | | matches | (empty) | |
| 100.00% | Filter word density | Target: ≤3% sentences with filter/hedge words | | filterCount | 1 | | hedgeCount | 0 | | narrationSentences | 113 | | filterMatches | | | hedgeMatches | (empty) | |
| 100.00% | Gibberish response detection | Target: ≤1% gibberish-like sentences (hard fail if a sentence exceeds 800 words) | | analyzedSentences | 149 | | gibberishSentences | 0 | | adjustedGibberishSentences | 0 | | longSentenceCount | 0 | | runOnParagraphCount | 0 | | giantParagraphCount | 0 | | wordSaladCount | 0 | | repetitionLoopCount | 0 | | controlTokenCount | 0 | | maxSentenceWordsSeen | 51 | | ratio | 0 | | matches | (empty) | |
| 100.00% | Markdown formatting overuse | Target: ≤5% words in markdown formatting | | markdownSpans | 0 | | markdownWords | 0 | | totalWords | 1651 | | ratio | 0 | | matches | (empty) | |
| 100.00% | Missing dialogue indicators (quotation marks) | Target: ≤10% speech attributions without quotation marks | | totalAttributions | 17 | | unquotedAttributions | 0 | | matches | (empty) | |
| 0.00% | Name drop frequency | Target: ≤1.0 per-name mentions per 100 words | | totalMentions | 52 | | wordCount | 1222 | | uniqueNames | 12 | | maxNameDensity | 1.96 | | worstName | "Quinn" | | maxWindowNameDensity | 5 | | worstWindowName | "Quinn" | | discoveredNames | | Quinn | 24 | | Camden | 1 | | Town | 1 | | Morris | 2 | | Constable | 1 | | Patel | 3 | | Fraser | 2 | | Serious | 1 | | Crimes | 1 | | Kowalski | 2 | | Silence | 1 | | Eva | 13 |
| | persons | | 0 | "Quinn" | | 1 | "Morris" | | 2 | "Constable" | | 3 | "Patel" | | 4 | "Fraser" | | 5 | "Kowalski" | | 6 | "Eva" |
| | places | | 0 | "Camden" | | 1 | "Town" | | 2 | "Serious" |
| | globalScore | 0.518 | | windowScore | 0 | |
| 38.06% | Narrator intent-glossing | Target: ≤2% narration sentences with intent-glossing patterns | | analyzedSentences | 67 | | glossingSentenceCount | 3 | | matches | | 0 | "seemed enormous freed from the press of commuters that would have given it scale" | | 1 | "looked like it belonged in a university l" | | 2 | "quite grief but was adjacent to it" |
| |
| 100.00% | "Not X but Y" pattern overuse | Target: ≤1 "not X but Y" per 1000 words | | totalMatches | 1 | | per1kWords | 0.606 | | wordCount | 1651 | | matches | | 0 | "not surprise exactly, but a rapid, private recalibration" |
| |
| 100.00% | Overuse of "that" (subordinate clause padding) | Target: ≤2% sentences with "that" clauses | | thatCount | 0 | | totalSentences | 149 | | matches | (empty) | |
| 100.00% | Paragraph length variance | Target: CV ≥0.5 for paragraph word counts | | totalParagraphs | 70 | | mean | 23.59 | | std | 22.24 | | cv | 0.943 | | sampleLengths | | 0 | 84 | | 1 | 59 | | 2 | 44 | | 3 | 44 | | 4 | 22 | | 5 | 1 | | 6 | 14 | | 7 | 46 | | 8 | 78 | | 9 | 3 | | 10 | 15 | | 11 | 38 | | 12 | 8 | | 13 | 3 | | 14 | 52 | | 15 | 33 | | 16 | 5 | | 17 | 17 | | 18 | 41 | | 19 | 6 | | 20 | 5 | | 21 | 7 | | 22 | 8 | | 23 | 83 | | 24 | 6 | | 25 | 10 | | 26 | 11 | | 27 | 3 | | 28 | 3 | | 29 | 66 | | 30 | 15 | | 31 | 66 | | 32 | 9 | | 33 | 20 | | 34 | 29 | | 35 | 1 | | 36 | 5 | | 37 | 34 | | 38 | 6 | | 39 | 17 | | 40 | 18 | | 41 | 5 | | 42 | 1 | | 43 | 39 | | 44 | 15 | | 45 | 4 | | 46 | 42 | | 47 | 12 | | 48 | 8 | | 49 | 10 |
| |
| 89.74% | Passive voice overuse | Target: ≤2% passive sentences | | passiveCount | 5 | | totalSentences | 113 | | matches | | 0 | "been shuttered" | | 1 | "being called" | | 2 | "was positioned" | | 3 | "been picked" | | 4 | "been removed" |
| |
| 73.62% | Past progressive (was/were + -ing) overuse | Target: ≤2% past progressive verbs | | pastProgressiveCount | 4 | | totalVerbs | 211 | | matches | | 0 | "was lying" | | 1 | "was talking" | | 2 | "was bulging" | | 3 | "was drawing" |
| |
| 0.00% | Em-dash & semicolon overuse | Target: ≤2% sentences with em-dashes/semicolons | | emDashCount | 13 | | semicolonCount | 1 | | flaggedSentences | 12 | | totalSentences | 149 | | ratio | 0.081 | | matches | | 0 | "Now she stood at the mouth of the service entrance to the old Camden Town station — the one that had been shuttered since 1924, according to the placard some heritage board had bolted to the wall and nobody ever read — and she watched the forensics team set up their lights below." | | 1 | "She stood, noted the time on her leather watch — 2:51 a.m." | | 2 | "— and descended the stairs." | | 3 | "His clothes — good quality, a wool overcoat, leather-soled shoes — were entirely unruffled, as though he'd lain down for a nap and simply not gotten up." | | 4 | "Not clenched — closed, as if he'd been holding something and the grip had relaxed in death." | | 5 | "The platform was dirty with decades of grime and old dust, and in that dust there were disturbances — footprints, multiple sets, some from the forensics team and some older." | | 6 | "Not graffiti — too deliberate, too evenly spaced." | | 7 | "The woman stopped when she saw the cordon tape and the assembled officers, and something crossed her face — not surprise exactly, but a rapid, private recalibration." | | 8 | "She was nine inches taller than this woman but she didn't use the height as a weapon — she simply held her ground and waited." | | 9 | "Quinn had learned, over eighteen years, to read the difference between performed emotion and real feeling, and this was real — a flash of something that wasn't quite grief but was adjacent to it." | | 10 | "Then she looked at Quinn, and in her eyes there was a calculation happening — the same kind Quinn recognised from mirrors, the interior arithmetic of someone deciding how much truth was safe." | | 11 | "\"Information. Substances. Tools.\" Eva's voice had steadied; she was on familiar ground now, academic footing." |
| |
| 100.00% | Purple prose (modifier overload) | Target: <4% adverbs, <2% -ly adverbs, no adj stacking | | wordCount | 1214 | | adjectiveStacks | 0 | | stackExamples | (empty) | | adverbCount | 46 | | adverbRatio | 0.03789126853377265 | | lyAdverbCount | 16 | | lyAdverbRatio | 0.013179571663920923 | |
| 100.00% | Repeated phrase echo | Target: ≤20% sentences with echoes (window: 2) | | totalSentences | 149 | | echoCount | 0 | | echoWords | (empty) | |
| 100.00% | Sentence length variance | Target: CV ≥0.4 for sentence word counts | | totalSentences | 149 | | mean | 11.08 | | std | 9.86 | | cv | 0.89 | | sampleLengths | | 0 | 31 | | 1 | 53 | | 2 | 16 | | 3 | 6 | | 4 | 20 | | 5 | 12 | | 6 | 5 | | 7 | 18 | | 8 | 21 | | 9 | 5 | | 10 | 11 | | 11 | 33 | | 12 | 6 | | 13 | 16 | | 14 | 1 | | 15 | 10 | | 16 | 4 | | 17 | 25 | | 18 | 8 | | 19 | 13 | | 20 | 19 | | 21 | 19 | | 22 | 8 | | 23 | 3 | | 24 | 2 | | 25 | 27 | | 26 | 3 | | 27 | 15 | | 28 | 25 | | 29 | 13 | | 30 | 6 | | 31 | 2 | | 32 | 3 | | 33 | 52 | | 34 | 6 | | 35 | 27 | | 36 | 2 | | 37 | 3 | | 38 | 11 | | 39 | 4 | | 40 | 2 | | 41 | 11 | | 42 | 5 | | 43 | 17 | | 44 | 8 | | 45 | 6 | | 46 | 5 | | 47 | 7 | | 48 | 8 | | 49 | 8 |
| |
| 64.43% | Sentence opener variety | Target: ≥60% unique sentence openers | | consecutiveRepeats | 14 | | diversityRatio | 0.4429530201342282 | | totalSentences | 149 | | uniqueOpeners | 66 | |
| 100.00% | Adverb-first sentence starts | Target: ≥3% sentences starting with an adverb | | adverbCount | 3 | | totalSentences | 92 | | matches | | 0 | "Then she looked at the" | | 1 | "Then she looked at Quinn," | | 2 | "Then she reached into her" |
| | ratio | 0.033 | |
| 98.26% | Pronoun-first sentence starts | Target: ≤30% sentences starting with a pronoun | | pronounCount | 28 | | totalSentences | 92 | | matches | | 0 | "She stood, noted the time" | | 1 | "She pushed the thought away." | | 2 | "He was young enough that" | | 3 | "He lifted the cordon tape" | | 4 | "His face was slack, his" | | 5 | "His clothes — good quality," | | 6 | "He was talking to a" | | 7 | "he said when she approached" | | 8 | "She turned away from him" | | 9 | "She didn't touch anything." | | 10 | "His left hand was closed." | | 11 | "She photographed it with her" | | 12 | "He materialised at her elbow" | | 13 | "She walked toward the tunnel" | | 14 | "She took out her torch" | | 15 | "She photographed them." | | 16 | "She was still looking at" | | 17 | "She had her badge out" | | 18 | "Her hand went to the" | | 19 | "It was something closer to" |
| | ratio | 0.304 | |
| 46.96% | Subject-first sentence starts | Target: ≤72% sentences starting with a subject | | subjectCount | 76 | | totalSentences | 92 | | matches | | 0 | "The call had come in" | | 1 | "The entrance was a steel" | | 2 | "Someone had burned the lock" | | 3 | "Quinn crouched and examined the" | | 4 | "She stood, noted the time" | | 5 | "The station had that particular" | | 6 | "Something that reminded her, in" | | 7 | "She pushed the thought away." | | 8 | "Detective Constable Patel met her" | | 9 | "He was young enough that" | | 10 | "He lifted the cordon tape" | | 11 | "The platform was enormous in" | | 12 | "The forensics lights turned everything" | | 13 | "Quinn walked the length of" | | 14 | "The body was positioned at" | | 15 | "The man was lying on" | | 16 | "His face was slack, his" | | 17 | "His clothes — good quality," | | 18 | "Patel tilted his head toward" | | 19 | "Fraser was a barrel-chested man" |
| | ratio | 0.826 | |
| 100.00% | Subordinate conjunction sentence starts | Target: ≥2% sentences starting with a subordinating conjunction | | subConjCount | 2 | | totalSentences | 92 | | matches | | 0 | "Now she stood at the" | | 1 | "— and descended the stairs." |
| | ratio | 0.022 | |
| 43.19% | Technical jargon density | Target: ≤6% sentences with technical-jargon patterns | | analyzedSentences | 43 | | technicalSentenceCount | 6 | | matches | | 0 | "The platform was enormous in the way that abandoned spaces always seemed enormous, freed from the press of commuters that would have given it scale." | | 1 | "His clothes — good quality, a wool overcoat, leather-soled shoes — were entirely unruffled, as though he'd lain down for a nap and simply not gotten up." | | 2 | "Fraser was a barrel-chested man from Serious Crimes who had a habit of reaching conclusions before the evidence did and then constructing a path backward." | | 3 | "The woman's green eyes were moving over the scene with a quality of observation that wasn't voyeurism and wasn't horror." | | 4 | "Quinn had learned, over eighteen years, to read the difference between performed emotion and real feeling, and this was real — a flash of something that wasn't …" | | 5 | "The composed, bloodless stillness of a man who had come here to buy something, or to sell it, or perhaps to take it from someone who hadn't wanted to give it up…" |
| |
| 100.00% | Useless dialogue additions | Target: ≤5% dialogue tags with trailing filler fragments | | totalTags | 23 | | uselessAdditionCount | 1 | | matches | | 0 | "Eva's voice had, academic footing" |
| |
| 100.00% | Dialogue tag variety (said vs. fancy) | Target: ≤10% fancy dialogue tags | | totalTags | 12 | | fancyCount | 0 | | fancyTags | (empty) | | dialogueSentences | 59 | | tagDensity | 0.203 | | leniency | 0.407 | | rawRatio | 0 | | effectiveRatio | 0 | |