| 100.00% | Adverbs in dialogue tags | Target: ≤10% dialogue tags with adverbs | | totalTags | 14 | | adverbTagCount | 1 | | adverbTags | | 0 | "Silas asked quietly [quietly]" |
| | dialogueSentences | 46 | | tagDensity | 0.304 | | leniency | 0.609 | | rawRatio | 0.071 | | effectiveRatio | 0.043 | |
| 86.34% | AI-ism adverb frequency | Target: <2% AI-ism adverbs (58 tracked) | | wordCount | 1098 | | totalAiIsmAdverbs | 3 | | found | | | highlights | | |
| 100.00% | AI-ism character names | Target: 0 AI-default names (16 tracked, −20% each) | | codexExemptions | | | found | (empty) | |
| 100.00% | AI-ism location names | Target: 0 AI-default location names (33 tracked, −20% each) | | codexExemptions | (empty) | | found | (empty) | |
| 36.25% | AI-ism word frequency | Target: <2% AI-ism words (290 tracked) | | wordCount | 1098 | | totalAiIsms | 14 | | found | | | highlights | | 0 | "flickered" | | 1 | "familiar" | | 2 | "flicked" | | 3 | "weight" | | 4 | "echoing" | | 5 | "silence" | | 6 | "unreadable" | | 7 | "charged" | | 8 | "unspoken" |
| |
| 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 | 93 | | matches | (empty) | |
| 100.00% | Filter word density | Target: ≤3% sentences with filter/hedge words | | filterCount | 0 | | hedgeCount | 0 | | narrationSentences | 93 | | filterMatches | (empty) | | hedgeMatches | (empty) | |
| 100.00% | Gibberish response detection | Target: ≤1% gibberish-like sentences (hard fail if a sentence exceeds 800 words) | | analyzedSentences | 125 | | gibberishSentences | 0 | | adjustedGibberishSentences | 0 | | longSentenceCount | 0 | | runOnParagraphCount | 0 | | giantParagraphCount | 0 | | wordSaladCount | 0 | | repetitionLoopCount | 0 | | controlTokenCount | 0 | | maxSentenceWordsSeen | 39 | | ratio | 0 | | matches | (empty) | |
| 100.00% | Markdown formatting overuse | Target: ≤5% words in markdown formatting | | markdownSpans | 6 | | markdownWords | 32 | | totalWords | 1089 | | ratio | 0.029 | | matches | | 0 | "clink" | | 1 | "get out now" | | 2 | "lets" | | 3 | "This is what happens when you trust the wrong people. This is what happens when you stay too long." | | 4 | "did" | | 5 | "Didn’t want to end up like you." |
| |
| 100.00% | Missing dialogue indicators (quotation marks) | Target: ≤10% speech attributions without quotation marks | | totalAttributions | 14 | | unquotedAttributions | 0 | | matches | (empty) | |
| 90.62% | Name drop frequency | Target: ≤1.0 per-name mentions per 100 words | | totalMentions | 31 | | wordCount | 842 | | uniqueNames | 13 | | maxNameDensity | 1.19 | | worstName | "Silas" | | maxWindowNameDensity | 2 | | worstWindowName | "Silas" | | discoveredNames | | Raven | 1 | | Nest | 1 | | Soho | 1 | | Carter | 1 | | Blackwood | 1 | | Silas | 10 | | Evan | 2 | | London | 1 | | Cardiff | 1 | | Eva | 1 | | Rory | 9 | | Prague | 1 | | Slid | 1 |
| | persons | | 0 | "Raven" | | 1 | "Carter" | | 2 | "Blackwood" | | 3 | "Silas" | | 4 | "Evan" | | 5 | "Eva" | | 6 | "Rory" |
| | places | | 0 | "Soho" | | 1 | "London" | | 2 | "Cardiff" | | 3 | "Prague" |
| | globalScore | 0.906 | | windowScore | 1 | |
| 100.00% | Narrator intent-glossing | Target: ≤2% narration sentences with intent-glossing patterns | | analyzedSentences | 62 | | glossingSentenceCount | 0 | | matches | (empty) | |
| 100.00% | "Not X but Y" pattern overuse | Target: ≤1 "not X but Y" per 1000 words | | totalMatches | 0 | | per1kWords | 0 | | wordCount | 1089 | | matches | (empty) | |
| 100.00% | Overuse of "that" (subordinate clause padding) | Target: ≤2% sentences with "that" clauses | | thatCount | 0 | | totalSentences | 125 | | matches | (empty) | |
| 100.00% | Paragraph length variance | Target: CV ≥0.5 for paragraph word counts | | totalParagraphs | 58 | | mean | 18.78 | | std | 15.61 | | cv | 0.831 | | sampleLengths | | 0 | 75 | | 1 | 51 | | 2 | 2 | | 3 | 26 | | 4 | 16 | | 5 | 25 | | 6 | 19 | | 7 | 42 | | 8 | 5 | | 9 | 5 | | 10 | 28 | | 11 | 33 | | 12 | 33 | | 13 | 6 | | 14 | 1 | | 15 | 7 | | 16 | 12 | | 17 | 27 | | 18 | 8 | | 19 | 1 | | 20 | 48 | | 21 | 6 | | 22 | 5 | | 23 | 5 | | 24 | 32 | | 25 | 18 | | 26 | 30 | | 27 | 4 | | 28 | 3 | | 29 | 4 | | 30 | 20 | | 31 | 45 | | 32 | 15 | | 33 | 17 | | 34 | 30 | | 35 | 4 | | 36 | 27 | | 37 | 2 | | 38 | 7 | | 39 | 10 | | 40 | 21 | | 41 | 9 | | 42 | 46 | | 43 | 9 | | 44 | 18 | | 45 | 41 | | 46 | 16 | | 47 | 8 | | 48 | 12 | | 49 | 47 |
| |
| 100.00% | Passive voice overuse | Target: ≤2% passive sentences | | passiveCount | 1 | | totalSentences | 93 | | matches | | |
| 100.00% | Past progressive (was/were + -ing) overuse | Target: ≤2% past progressive verbs | | pastProgressiveCount | 1 | | totalVerbs | 150 | | matches | | |
| 0.00% | Em-dash & semicolon overuse | Target: ≤2% sentences with em-dashes/semicolons | | emDashCount | 9 | | semicolonCount | 0 | | flaggedSentences | 7 | | totalSentences | 125 | | ratio | 0.056 | | matches | | 0 | "She’d been here a hundred times before—delivering takeaway orders, slipping in for a quick pint after a long shift—but tonight felt different." | | 1 | "Then—" | | 2 | "Just studied her—the way her black hair fell straight past her shoulders, the way her left wrist twitched when she was lying." | | 3 | "But the weight of the last few months pressed down on her—Evan’s voice still echoing in her skull, the way London had swallowed her whole after Cardiff." | | 4 | "She looked at him then—really looked." | | 5 | "She’d known it since she was fifteen, sneaking into his flat after school, listening to the stories he’d tell her—half-truths wrapped in cautionary tales." | | 6 | "The bar hummed around them—laughter from a corner booth, the clink of glasses, the low murmur of conversation." |
| |
| 100.00% | Purple prose (modifier overload) | Target: <4% adverbs, <2% -ly adverbs, no adj stacking | | wordCount | 851 | | adjectiveStacks | 0 | | stackExamples | (empty) | | adverbCount | 29 | | adverbRatio | 0.03407755581668625 | | lyAdverbCount | 5 | | lyAdverbRatio | 0.005875440658049354 | |
| 100.00% | Repeated phrase echo | Target: ≤20% sentences with echoes (window: 2) | | totalSentences | 125 | | echoCount | 0 | | echoWords | (empty) | |
| 100.00% | Sentence length variance | Target: CV ≥0.4 for sentence word counts | | totalSentences | 125 | | mean | 8.71 | | std | 6.12 | | cv | 0.703 | | sampleLengths | | 0 | 18 | | 1 | 21 | | 2 | 22 | | 3 | 14 | | 4 | 16 | | 5 | 14 | | 6 | 15 | | 7 | 5 | | 8 | 1 | | 9 | 2 | | 10 | 14 | | 11 | 6 | | 12 | 6 | | 13 | 9 | | 14 | 7 | | 15 | 9 | | 16 | 8 | | 17 | 8 | | 18 | 9 | | 19 | 10 | | 20 | 3 | | 21 | 22 | | 22 | 8 | | 23 | 9 | | 24 | 5 | | 25 | 5 | | 26 | 2 | | 27 | 10 | | 28 | 15 | | 29 | 1 | | 30 | 4 | | 31 | 27 | | 32 | 2 | | 33 | 12 | | 34 | 11 | | 35 | 10 | | 36 | 6 | | 37 | 1 | | 38 | 7 | | 39 | 10 | | 40 | 2 | | 41 | 7 | | 42 | 11 | | 43 | 9 | | 44 | 5 | | 45 | 3 | | 46 | 1 | | 47 | 13 | | 48 | 20 | | 49 | 15 |
| |
| 50.40% | Sentence opener variety | Target: ≥60% unique sentence openers | | consecutiveRepeats | 7 | | diversityRatio | 0.336 | | totalSentences | 125 | | uniqueOpeners | 42 | |
| 100.00% | Adverb-first sentence starts | Target: ≥3% sentences starting with an adverb | | adverbCount | 4 | | totalSentences | 82 | | matches | | 0 | "Just studied her—the way her" | | 1 | "Then, the ghost of a" | | 2 | "Instead, she reached for her" | | 3 | "Then he turned, limping slightly" |
| | ratio | 0.049 | |
| 68.78% | Pronoun-first sentence starts | Target: ≤30% sentences starting with a pronoun | | pronounCount | 31 | | totalSentences | 82 | | matches | | 0 | "She’d been here a hundred" | | 1 | "She slid onto a stool" | | 2 | "His back was to her," | | 3 | "His voice cut through the" | | 4 | "She hadn’t even seen him" | | 5 | "she said, keeping her tone" | | 6 | "His hazel eyes flicked up," | | 7 | "She exhaled through her nose," | | 8 | "He didn’t smile." | | 9 | "She’d been seventeen the last" | | 10 | "He reached beneath the counter," | | 11 | "She should’ve said no." | | 12 | "She dragged her thumb along" | | 13 | "She corrected him automatically" | | 14 | "She didn’t say any of" | | 15 | "He reached for the bottle" | | 16 | "She looked at him then—really" | | 17 | "His laugh was a dry," | | 18 | "She’d known it since she" | | 19 | "His fingers tightened around the" |
| | ratio | 0.378 | |
| 63.66% | Subject-first sentence starts | Target: ≤72% sentences starting with a subject | | subjectCount | 65 | | totalSentences | 82 | | matches | | 0 | "The neon glow of the" | | 1 | "Rory Carter pushed through the" | | 2 | "She’d been here a hundred" | | 3 | "The air hummed with something" | | 4 | "She slid onto a stool" | | 5 | "Silas Blackwood stood at the" | | 6 | "His back was to her," | | 7 | "His voice cut through the" | | 8 | "Rory’s fingers stilled on the" | | 9 | "She hadn’t even seen him" | | 10 | "she said, keeping her tone" | | 11 | "Silas set the glass down" | | 12 | "His hazel eyes flicked up," | | 13 | "She exhaled through her nose," | | 14 | "He didn’t smile." | | 15 | "The crescent scar there, pale" | | 16 | "She’d been seventeen the last" | | 17 | "He reached beneath the counter," | | 18 | "She should’ve said no." | | 19 | "Silas poured two fingers into" |
| | ratio | 0.793 | |
| 100.00% | Subordinate conjunction sentence starts | Target: ≥2% sentences starting with a subordinating conjunction | | subConjCount | 3 | | totalSentences | 82 | | matches | | 0 | "Because Evan had split her" | | 1 | "Because Eva had shown up" | | 2 | "Because Rory had spent years" |
| | ratio | 0.037 | |
| 100.00% | Technical jargon density | Target: ≤6% sentences with technical-jargon patterns | | analyzedSentences | 34 | | technicalSentenceCount | 0 | | matches | (empty) | |
| 89.29% | Useless dialogue additions | Target: ≤5% dialogue tags with trailing filler fragments | | totalTags | 14 | | uselessAdditionCount | 1 | | matches | | 0 | "He leaned in, voice dropping" |
| |
| 100.00% | Dialogue tag variety (said vs. fancy) | Target: ≤10% fancy dialogue tags | | totalTags | 8 | | fancyCount | 2 | | fancyTags | | 0 | "She corrected (correct)" | | 1 | "she lied (lie)" |
| | dialogueSentences | 46 | | tagDensity | 0.174 | | leniency | 0.348 | | rawRatio | 0.25 | | effectiveRatio | 0.087 | |