| 100.00% | Adverbs in dialogue tags | Target: ≤10% dialogue tags with adverbs | | totalTags | 21 | | adverbTagCount | 0 | | adverbTags | (empty) | | dialogueSentences | 63 | | tagDensity | 0.333 | | leniency | 0.667 | | rawRatio | 0 | | effectiveRatio | 0 | |
| 87.01% | AI-ism adverb frequency | Target: <2% AI-ism adverbs (58 tracked) | | wordCount | 1540 | | totalAiIsmAdverbs | 4 | | found | | | highlights | | 0 | "completely" | | 1 | "very" | | 2 | "really" | | 3 | "gently" |
| |
| 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) | |
| 70.78% | AI-ism word frequency | Target: <2% AI-ism words (290 tracked) | | wordCount | 1540 | | totalAiIsms | 9 | | found | | | highlights | | 0 | "perfect" | | 1 | "synthetic" | | 2 | "weight" | | 3 | "echoed" | | 4 | "familiar" | | 5 | "crystalline" | | 6 | "potential" | | 7 | "shattered" |
| |
| 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 | 181 | | matches | (empty) | |
| 95.50% | Filter word density | Target: ≤3% sentences with filter/hedge words | | filterCount | 4 | | hedgeCount | 2 | | narrationSentences | 181 | | filterMatches | | | hedgeMatches | | |
| 100.00% | Gibberish response detection | Target: ≤1% gibberish-like sentences (hard fail if a sentence exceeds 800 words) | | analyzedSentences | 221 | | gibberishSentences | 0 | | adjustedGibberishSentences | 0 | | longSentenceCount | 0 | | runOnParagraphCount | 0 | | giantParagraphCount | 0 | | wordSaladCount | 0 | | repetitionLoopCount | 0 | | controlTokenCount | 0 | | maxSentenceWordsSeen | 26 | | ratio | 0 | | matches | (empty) | |
| 100.00% | Markdown formatting overuse | Target: ≤5% words in markdown formatting | | markdownSpans | 2 | | markdownWords | 16 | | totalWords | 1528 | | ratio | 0.01 | | matches | | 0 | "Essence of Forgotten Names. Bottled Thursday's Child. London Rain (1888 vintage)." | | 1 | "One Police Detective, Gently Used." |
| |
| 100.00% | Missing dialogue indicators (quotation marks) | Target: ≤10% speech attributions without quotation marks | | totalAttributions | 11 | | unquotedAttributions | 0 | | matches | (empty) | |
| 57.41% | Name drop frequency | Target: ≤1.0 per-name mentions per 100 words | | totalMentions | 68 | | wordCount | 1134 | | uniqueNames | 25 | | maxNameDensity | 1.85 | | worstName | "Patel" | | maxWindowNameDensity | 3 | | worstWindowName | "Patel" | | discoveredNames | | Harlow | 1 | | Quinn | 15 | | Camden | 2 | | Market | 1 | | Tube | 1 | | Patel | 21 | | Victorian | 1 | | Underground | 1 | | Row | 1 | | Italian | 1 | | Morris | 3 | | Forgotten | 1 | | Names | 1 | | Thursday | 1 | | Child | 1 | | Rain | 1 | | Deeper | 1 | | Gold | 1 | | Not-Patel | 6 | | Worried | 1 | | Occult | 1 | | Supplies | 1 | | Police | 1 | | Detective | 2 | | Gently | 1 |
| | persons | | 0 | "Harlow" | | 1 | "Quinn" | | 2 | "Patel" | | 3 | "Morris" | | 4 | "Thursday" | | 5 | "Not-Patel" |
| | places | (empty) | | globalScore | 0.574 | | windowScore | 0.667 | |
| 100.00% | Narrator intent-glossing | Target: ≤2% narration sentences with intent-glossing patterns | | analyzedSentences | 87 | | glossingSentenceCount | 1 | | matches | | |
| 100.00% | "Not X but Y" pattern overuse | Target: ≤1 "not X but Y" per 1000 words | | totalMatches | 0 | | per1kWords | 0 | | wordCount | 1528 | | matches | (empty) | |
| 100.00% | Overuse of "that" (subordinate clause padding) | Target: ≤2% sentences with "that" clauses | | thatCount | 0 | | totalSentences | 221 | | matches | (empty) | |
| 100.00% | Paragraph length variance | Target: CV ≥0.5 for paragraph word counts | | totalParagraphs | 82 | | mean | 18.63 | | std | 12.63 | | cv | 0.678 | | sampleLengths | | 0 | 8 | | 1 | 50 | | 2 | 11 | | 3 | 32 | | 4 | 4 | | 5 | 24 | | 6 | 33 | | 7 | 16 | | 8 | 32 | | 9 | 8 | | 10 | 4 | | 11 | 28 | | 12 | 6 | | 13 | 14 | | 14 | 43 | | 15 | 12 | | 16 | 35 | | 17 | 4 | | 18 | 24 | | 19 | 6 | | 20 | 5 | | 21 | 15 | | 22 | 29 | | 23 | 4 | | 24 | 2 | | 25 | 37 | | 26 | 8 | | 27 | 3 | | 28 | 25 | | 29 | 2 | | 30 | 12 | | 31 | 16 | | 32 | 7 | | 33 | 9 | | 34 | 31 | | 35 | 7 | | 36 | 30 | | 37 | 45 | | 38 | 9 | | 39 | 41 | | 40 | 19 | | 41 | 1 | | 42 | 17 | | 43 | 11 | | 44 | 7 | | 45 | 4 | | 46 | 18 | | 47 | 27 | | 48 | 32 | | 49 | 34 |
| |
| 100.00% | Passive voice overuse | Target: ≤2% passive sentences | | passiveCount | 2 | | totalSentences | 181 | | matches | | 0 | "were arranged" | | 1 | "was listed" |
| |
| 100.00% | Past progressive (was/were + -ing) overuse | Target: ≤2% past progressive verbs | | pastProgressiveCount | 1 | | totalVerbs | 198 | | matches | | |
| 100.00% | Em-dash & semicolon overuse | Target: ≤2% sentences with em-dashes/semicolons | | emDashCount | 0 | | semicolonCount | 0 | | flaggedSentences | 0 | | totalSentences | 221 | | ratio | 0 | | matches | (empty) | |
| 100.00% | Purple prose (modifier overload) | Target: <4% adverbs, <2% -ly adverbs, no adj stacking | | wordCount | 1144 | | adjectiveStacks | 0 | | stackExamples | (empty) | | adverbCount | 31 | | adverbRatio | 0.027097902097902096 | | lyAdverbCount | 10 | | lyAdverbRatio | 0.008741258741258742 | |
| 100.00% | Repeated phrase echo | Target: ≤20% sentences with echoes (window: 2) | | totalSentences | 221 | | echoCount | 0 | | echoWords | (empty) | |
| 100.00% | Sentence length variance | Target: CV ≥0.4 for sentence word counts | | totalSentences | 221 | | mean | 6.91 | | std | 4.69 | | cv | 0.678 | | sampleLengths | | 0 | 8 | | 1 | 16 | | 2 | 24 | | 3 | 10 | | 4 | 8 | | 5 | 3 | | 6 | 12 | | 7 | 6 | | 8 | 7 | | 9 | 7 | | 10 | 4 | | 11 | 17 | | 12 | 7 | | 13 | 12 | | 14 | 4 | | 15 | 2 | | 16 | 3 | | 17 | 12 | | 18 | 4 | | 19 | 12 | | 20 | 5 | | 21 | 8 | | 22 | 2 | | 23 | 2 | | 24 | 4 | | 25 | 6 | | 26 | 5 | | 27 | 8 | | 28 | 4 | | 29 | 6 | | 30 | 2 | | 31 | 6 | | 32 | 3 | | 33 | 7 | | 34 | 2 | | 35 | 2 | | 36 | 6 | | 37 | 8 | | 38 | 6 | | 39 | 6 | | 40 | 10 | | 41 | 12 | | 42 | 15 | | 43 | 6 | | 44 | 6 | | 45 | 11 | | 46 | 5 | | 47 | 2 | | 48 | 2 | | 49 | 12 |
| |
| 73.00% | Sentence opener variety | Target: ≥60% unique sentence openers | | consecutiveRepeats | 22 | | diversityRatio | 0.497737556561086 | | totalSentences | 221 | | uniqueOpeners | 110 | |
| 49.38% | Adverb-first sentence starts | Target: ≥3% sentences starting with an adverb | | adverbCount | 2 | | totalSentences | 135 | | matches | | 0 | "Too many teeth." | | 1 | "All wearing faces that didn't" |
| | ratio | 0.015 | |
| 100.00% | Pronoun-first sentence starts | Target: ≤30% sentences starting with a pronoun | | pronounCount | 18 | | totalSentences | 135 | | matches | | 0 | "She'd worked the Underground for" | | 1 | "She checked the label." | | 2 | "She lifted the corpse's left" | | 3 | "She rubbed her thumb across" | | 4 | "She'd seen enough corpses to" | | 5 | "She stood, brushed dust from" | | 6 | "She walked the length of" | | 7 | "She gestured toward the market" | | 8 | "She'd been underground during power" | | 9 | "She hoisted herself onto the" | | 10 | "Her hand found her radio." | | 11 | "Its face was still frozen" | | 12 | "She gripped the compass tighter." | | 13 | "He reached for the ledger." | | 14 | "Her radio crackled to life." | | 15 | "She burst onto the street," | | 16 | "She glanced back at the" | | 17 | "He gestured at the brick" |
| | ratio | 0.133 | |
| 82.22% | Subject-first sentence starts | Target: ≤72% sentences starting with a subject | | subjectCount | 102 | | totalSentences | 135 | | matches | | 0 | "The body shouldn't have been" | | 1 | "Detective Harlow Quinn ducked under" | | 2 | "The abandoned Tube station stretched" | | 3 | "DS Patel called out" | | 4 | "Quinn's torch beam swept across" | | 5 | "The platform curved away into" | | 6 | "She'd worked the Underground for" | | 7 | "Patel consulted his notebook" | | 8 | "The corpse lay sprawled across" | | 9 | "Quinn crouched closer." | | 10 | "Quinn pulled on latex gloves." | | 11 | "The suit fabric felt wrong" | | 12 | "She checked the label." | | 13 | "Savile Row, but no tailor's" | | 14 | "The stitching was perfect, machine-precise." | | 15 | "She lifted the corpse's left" | | 16 | "She rubbed her thumb across" | | 17 | "Patel shifted his weight" | | 18 | "Quinn's torch beam found it" | | 19 | "Medieval-looking signs hung above them," |
| | ratio | 0.756 | |
| 37.04% | Subordinate conjunction sentence starts | Target: ≥2% sentences starting with a subordinating conjunction | | subConjCount | 1 | | totalSentences | 135 | | matches | | | ratio | 0.007 | |
| 90.59% | Technical jargon density | Target: ≤6% sentences with technical-jargon patterns | | analyzedSentences | 41 | | technicalSentenceCount | 3 | | matches | | 0 | "The bottles sang in their shelves, a crystalline harmony that made her teeth ache." | | 1 | "But her palm tingled with phantom heat, and when she closed her eyes, she could still see the needle pointing toward things that wore human faces." | | 2 | "But in her jacket pocket, her fingers found something that hadn't been there before." |
| |
| 100.00% | Useless dialogue additions | Target: ≤5% dialogue tags with trailing filler fragments | | totalTags | 21 | | uselessAdditionCount | 0 | | matches | (empty) | |
| 100.00% | Dialogue tag variety (said vs. fancy) | Target: ≤10% fancy dialogue tags | | totalTags | 5 | | fancyCount | 2 | | fancyTags | | 0 | "Patel suggested (suggest)" | | 1 | "Patel suggested (suggest)" |
| | dialogueSentences | 63 | | tagDensity | 0.079 | | leniency | 0.159 | | rawRatio | 0.4 | | effectiveRatio | 0.063 | |