| 100.00% | Adverbs in dialogue tags | Target: ≤10% dialogue tags with adverbs | | totalTags | 10 | | adverbTagCount | 0 | | adverbTags | (empty) | | dialogueSentences | 30 | | tagDensity | 0.333 | | leniency | 0.667 | | rawRatio | 0 | | effectiveRatio | 0 | |
| 86.14% | AI-ism adverb frequency | Target: <2% AI-ism adverbs (58 tracked) | | wordCount | 1082 | | totalAiIsmAdverbs | 3 | | found | | | highlights | | 0 | "slowly" | | 1 | "slightly" | | 2 | "carefully" |
| |
| 100.00% | AI-ism character names | Target: 0 AI-default names (17 tracked, −20% each) | | codexExemptions | (empty) | | found | (empty) | |
| 100.00% | AI-ism location names | Target: 0 AI-default location names (33 tracked, −20% each) | | codexExemptions | (empty) | | found | (empty) | |
| 95.38% | AI-ism word frequency | Target: <2% AI-ism words (290 tracked) | | wordCount | 1082 | | totalAiIsms | 1 | | found | | | highlights | | |
| 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 | 67 | | matches | (empty) | |
| 100.00% | Filter word density | Target: ≤3% sentences with filter/hedge words | | filterCount | 1 | | hedgeCount | 0 | | narrationSentences | 67 | | filterMatches | | | hedgeMatches | (empty) | |
| 100.00% | Gibberish response detection | Target: ≤1% gibberish-like sentences (hard fail if a sentence exceeds 800 words) | | analyzedSentences | 87 | | gibberishSentences | 0 | | adjustedGibberishSentences | 0 | | longSentenceCount | 0 | | runOnParagraphCount | 0 | | giantParagraphCount | 0 | | wordSaladCount | 0 | | repetitionLoopCount | 0 | | controlTokenCount | 0 | | maxSentenceWordsSeen | 46 | | ratio | 0 | | matches | (empty) | |
| 100.00% | Markdown formatting overuse | Target: ≤5% words in markdown formatting | | markdownSpans | 0 | | markdownWords | 0 | | totalWords | 1084 | | ratio | 0 | | matches | (empty) | |
| 100.00% | Missing dialogue indicators (quotation marks) | Target: ≤10% speech attributions without quotation marks | | totalAttributions | 7 | | unquotedAttributions | 0 | | matches | (empty) | |
| 99.75% | Name drop frequency | Target: ≤1.0 per-name mentions per 100 words | | totalMentions | 21 | | wordCount | 796 | | uniqueNames | 7 | | maxNameDensity | 1.01 | | worstName | "Peters" | | maxWindowNameDensity | 1.5 | | worstWindowName | "Quinn" | | discoveredNames | | Tube | 1 | | Quinn | 7 | | Peters | 8 | | Younger | 1 | | Ashworth | 2 | | Bermondsey | 1 | | Morris | 1 |
| | persons | | 0 | "Quinn" | | 1 | "Peters" | | 2 | "Ashworth" | | 3 | "Morris" |
| | places | | | globalScore | 0.997 | | windowScore | 1 | |
| 100.00% | Narrator intent-glossing | Target: ≤2% narration sentences with intent-glossing patterns | | analyzedSentences | 40 | | glossingSentenceCount | 0 | | matches | (empty) | |
| 0.00% | "Not X but Y" pattern overuse | Target: ≤1 "not X but Y" per 1000 words | | totalMatches | 3 | | per1kWords | 2.768 | | wordCount | 1084 | | matches | | 0 | "not the usual cocktail of blood and forensic spray, but something older" | | 1 | "not graffiti, not gang tags, but something deliberate" | | 2 | "not gang tags, but something deliberate" |
| |
| 100.00% | Overuse of "that" (subordinate clause padding) | Target: ≤2% sentences with "that" clauses | | thatCount | 1 | | totalSentences | 87 | | matches | | |
| 100.00% | Paragraph length variance | Target: CV ≥0.5 for paragraph word counts | | totalParagraphs | 35 | | mean | 30.97 | | std | 24.78 | | cv | 0.8 | | sampleLengths | | 0 | 47 | | 1 | 20 | | 2 | 60 | | 3 | 46 | | 4 | 6 | | 5 | 39 | | 6 | 50 | | 7 | 30 | | 8 | 5 | | 9 | 20 | | 10 | 86 | | 11 | 17 | | 12 | 5 | | 13 | 3 | | 14 | 74 | | 15 | 41 | | 16 | 36 | | 17 | 12 | | 18 | 4 | | 19 | 53 | | 20 | 13 | | 21 | 7 | | 22 | 1 | | 23 | 12 | | 24 | 88 | | 25 | 7 | | 26 | 44 | | 27 | 19 | | 28 | 31 | | 29 | 12 | | 30 | 41 | | 31 | 6 | | 32 | 28 | | 33 | 81 | | 34 | 40 |
| |
| 73.84% | Passive voice overuse | Target: ≤2% passive sentences | | passiveCount | 6 | | totalSentences | 67 | | matches | | 0 | "been opened" | | 1 | "been disturbed" | | 2 | "been placed" | | 3 | "were drawn" | | 4 | "been made" | | 5 | "been removed" | | 6 | "been cleared" |
| |
| 100.00% | Past progressive (was/were + -ing) overuse | Target: ≤2% past progressive verbs | | pastProgressiveCount | 0 | | totalVerbs | 112 | | matches | (empty) | |
| 0.00% | Em-dash & semicolon overuse | Target: ≤2% sentences with em-dashes/semicolons | | emDashCount | 6 | | semicolonCount | 0 | | flaggedSentences | 5 | | totalSentences | 87 | | ratio | 0.057 | | matches | | 0 | "The yellow tape had gone up around half the abandoned Tube platform before Quinn arrived, and the scene already smelled wrong — not the usual cocktail of blood and forensic spray, but something older." | | 1 | "Chalk markings covered the tiles in concentric rings — not graffiti, not gang tags, but something deliberate." | | 2 | "Her torch swept along the platform wall and found what she was already looking for — scratch marks at hip height, dozens of them, concentrated around a section of tile near the far arch." | | 3 | "The dust lay thick and even across the platform except for two things: a worn path cutting in from the southern arch — foot traffic, recent, multiple visitors — and a small clean circle of tile roughly a metre from Ashworth's right hand, where something had been removed." | | 4 | "The booth itself had been cleared of decades of detritus — she could see the clean rectangle of the counter beneath a film of recently settled grime, as though someone had wiped it down weeks ago and the station had since begun to reclaim it." |
| |
| 100.00% | Purple prose (modifier overload) | Target: <4% adverbs, <2% -ly adverbs, no adj stacking | | wordCount | 465 | | adjectiveStacks | 0 | | stackExamples | (empty) | | adverbCount | 9 | | adverbRatio | 0.01935483870967742 | | lyAdverbCount | 4 | | lyAdverbRatio | 0.008602150537634409 | |
| 100.00% | Repeated phrase echo | Target: ≤20% sentences with echoes (window: 2) | | totalSentences | 87 | | echoCount | 0 | | echoWords | (empty) | |
| 100.00% | Sentence length variance | Target: CV ≥0.4 for sentence word counts | | totalSentences | 87 | | mean | 12.46 | | std | 10.94 | | cv | 0.878 | | sampleLengths | | 0 | 34 | | 1 | 1 | | 2 | 12 | | 3 | 20 | | 4 | 17 | | 5 | 1 | | 6 | 11 | | 7 | 25 | | 8 | 3 | | 9 | 3 | | 10 | 20 | | 11 | 21 | | 12 | 5 | | 13 | 6 | | 14 | 32 | | 15 | 7 | | 16 | 13 | | 17 | 5 | | 18 | 10 | | 19 | 22 | | 20 | 11 | | 21 | 19 | | 22 | 5 | | 23 | 12 | | 24 | 8 | | 25 | 2 | | 26 | 34 | | 27 | 2 | | 28 | 1 | | 29 | 3 | | 30 | 35 | | 31 | 1 | | 32 | 8 | | 33 | 17 | | 34 | 5 | | 35 | 3 | | 36 | 3 | | 37 | 23 | | 38 | 48 | | 39 | 11 | | 40 | 5 | | 41 | 14 | | 42 | 1 | | 43 | 4 | | 44 | 6 | | 45 | 9 | | 46 | 27 | | 47 | 8 | | 48 | 4 | | 49 | 4 |
| |
| 94.25% | Sentence opener variety | Target: ≥60% unique sentence openers | | consecutiveRepeats | 3 | | diversityRatio | 0.5862068965517241 | | totalSentences | 87 | | uniqueOpeners | 51 | |
| 61.73% | Adverb-first sentence starts | Target: ≥3% sentences starting with an adverb | | adverbCount | 1 | | totalSentences | 54 | | matches | | | ratio | 0.019 | |
| 100.00% | Pronoun-first sentence starts | Target: ≤30% sentences starting with a pronoun | | pronounCount | 14 | | totalSentences | 54 | | matches | | 0 | "She ducked under the tape" | | 1 | "He gestured at the body" | | 2 | "They had not been disturbed" | | 3 | "Her torch swept along the" | | 4 | "She walked to it, and" | | 5 | "She hunkered down and examined" | | 6 | "She drew her finger parallel" | | 7 | "He had the good grace" | | 8 | "She moved back toward the" | | 9 | "She turned and found Peters" | | 10 | "She looked back at the" | | 11 | "She straightened, rolling her left" | | 12 | "She photographed each one separately" | | 13 | "She had seen that script" |
| | ratio | 0.259 | |
| 89.63% | Subject-first sentence starts | Target: ≤72% sentences starting with a subject | | subjectCount | 40 | | totalSentences | 54 | | matches | | 0 | "The yellow tape had gone" | | 1 | "She ducked under the tape" | | 2 | "Chalk markings covered the tiles" | | 3 | "The kind of thing that" | | 4 | "A single body lay at" | | 5 | "DS Peters materialised from a" | | 6 | "He gestured at the body" | | 7 | "Quinn crouched at the edge" | | 8 | "The lines were clean, unsmudged." | | 9 | "They had not been disturbed" | | 10 | "Peters said, reading from the" | | 11 | "Her torch swept along the" | | 12 | "The same spot." | | 13 | "She walked to it, and" | | 14 | "Peters typed something." | | 15 | "Quinn moved back toward the" | | 16 | "The dust lay thick and" | | 17 | "She hunkered down and examined" | | 18 | "The tile was lighter here." | | 19 | "An object had rested in" |
| | ratio | 0.741 | |
| 92.59% | Subordinate conjunction sentence starts | Target: ≥2% sentences starting with a subordinating conjunction | | subConjCount | 1 | | totalSentences | 54 | | matches | | 0 | "Either he had walked calmly" |
| | ratio | 0.019 | |
| 60.44% | Technical jargon density | Target: ≤6% sentences with technical-jargon patterns | | analyzedSentences | 26 | | technicalSentenceCount | 3 | | matches | | 0 | "A single body lay at the centre of the outermost ring, face-up, arms arranged with the careful symmetry of someone who had taken their time." | | 1 | "Younger than her by a decade, with the rosy-cheeked confidence of someone who had never had a case they couldn't close." | | 2 | "She walked to it, and the marks resolved into something that had been made by many hands over many visits: a groove worn into the mortar, roughly the size and s…" |
| |
| 75.00% | Useless dialogue additions | Target: ≤5% dialogue tags with trailing filler fragments | | totalTags | 10 | | uselessAdditionCount | 1 | | matches | | 0 | "DS Peters materialised, a tablet tucked under his arm" |
| |
| 100.00% | Dialogue tag variety (said vs. fancy) | Target: ≤10% fancy dialogue tags | | totalTags | 4 | | fancyCount | 0 | | fancyTags | (empty) | | dialogueSentences | 30 | | tagDensity | 0.133 | | leniency | 0.267 | | rawRatio | 0 | | effectiveRatio | 0 | |