| 100.00% | Adverbs in dialogue tags | Target: ≤10% dialogue tags with adverbs | |
| 100.00% | AI-ism adverb frequency | Target: <2% AI-ism adverbs (58 tracked) | | wordCount | 2462 | | totalAiIsmAdverbs | 0 | | found | (empty) | | highlights | (empty) | |
| 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) | |
| 81.72% | AI-ism word frequency | Target: <2% AI-ism words (290 tracked) | | wordCount | 2462 | | totalAiIsms | 9 | | found | | | highlights | | 0 | "footsteps" | | 1 | "flicked" | | 2 | "vibrated" | | 3 | "weight" | | 4 | "traced" | | 5 | "velvet" | | 6 | "could feel" |
| |
| 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 | 1 | | narrationSentences | 271 | | matches | | |
| 100.00% | Filter word density | Target: ≤3% sentences with filter/hedge words | | filterCount | 3 | | hedgeCount | 2 | | narrationSentences | 271 | | filterMatches | | | hedgeMatches | | |
| 100.00% | Gibberish response detection | Target: ≤1% gibberish-like sentences (hard fail if a sentence exceeds 800 words) | | analyzedSentences | 271 | | gibberishSentences | 0 | | adjustedGibberishSentences | 0 | | longSentenceCount | 0 | | runOnParagraphCount | 0 | | giantParagraphCount | 0 | | wordSaladCount | 0 | | repetitionLoopCount | 0 | | controlTokenCount | 0 | | maxSentenceWordsSeen | 38 | | ratio | 0 | | matches | (empty) | |
| 100.00% | Markdown formatting overuse | Target: ≤5% words in markdown formatting | | markdownSpans | 0 | | markdownWords | 0 | | totalWords | 2459 | | ratio | 0 | | matches | (empty) | |
| 41.67% | Missing dialogue indicators (quotation marks) | Target: ≤10% speech attributions without quotation marks | | totalAttributions | 3 | | unquotedAttributions | 1 | | matches | | |
| 100.00% | Name drop frequency | Target: ≤1.0 per-name mentions per 100 words | | totalMentions | 67 | | wordCount | 2455 | | uniqueNames | 24 | | maxNameDensity | 0.86 | | worstName | "Harlow" | | maxWindowNameDensity | 2 | | worstWindowName | "Harlow" | | discoveredNames | | Camden | 3 | | Lock | 1 | | Police | 2 | | Control | 1 | | Harlow | 21 | | Victorian | 1 | | Morris | 1 | | Berwick | 1 | | Finest | 1 | | Town | 1 | | Eastbound | 1 | | Herrera | 1 | | Saint | 1 | | Christopher | 1 | | Tomás | 7 | | Sevilla | 1 | | Quinn | 2 | | Wardens | 4 | | Warden | 2 | | Close | 1 | | Bone | 3 | | Token | 3 | | Stalls | 3 | | You | 4 |
| | persons | | 0 | "Harlow" | | 1 | "Morris" | | 2 | "Herrera" | | 3 | "Saint" | | 4 | "Christopher" | | 5 | "Tomás" | | 6 | "Quinn" | | 7 | "Wardens" | | 8 | "Warden" | | 9 | "Token" | | 10 | "Stalls" | | 11 | "You" |
| | places | | 0 | "Camden" | | 1 | "Control" | | 2 | "Berwick" | | 3 | "Town" | | 4 | "Eastbound" |
| | globalScore | 1 | | windowScore | 1 | |
| 93.18% | Narrator intent-glossing | Target: ≤2% narration sentences with intent-glossing patterns | | analyzedSentences | 176 | | glossingSentenceCount | 4 | | matches | | 0 | "tasted like a battery" | | 1 | "something like a smile found one corner of h" | | 2 | "sounded like old brakes catching on metal" | | 3 | "smelled like the inside of a mausoleum jus" |
| |
| 100.00% | "Not X but Y" pattern overuse | Target: ≤1 "not X but Y" per 1000 words | | totalMatches | 1 | | per1kWords | 0.407 | | wordCount | 2459 | | matches | | 0 | "not with heat, but with force like a wave hitting the pier" |
| |
| 100.00% | Overuse of "that" (subordinate clause padding) | Target: ≤2% sentences with "that" clauses | | thatCount | 1 | | totalSentences | 271 | | matches | | |
| 100.00% | Paragraph length variance | Target: CV ≥0.5 for paragraph word counts | | totalParagraphs | 64 | | mean | 38.42 | | std | 25.58 | | cv | 0.666 | | sampleLengths | | 0 | 28 | | 1 | 75 | | 2 | 50 | | 3 | 51 | | 4 | 58 | | 5 | 10 | | 6 | 21 | | 7 | 91 | | 8 | 39 | | 9 | 65 | | 10 | 3 | | 11 | 16 | | 12 | 13 | | 13 | 78 | | 14 | 49 | | 15 | 59 | | 16 | 24 | | 17 | 21 | | 18 | 83 | | 19 | 92 | | 20 | 65 | | 21 | 83 | | 22 | 61 | | 23 | 1 | | 24 | 60 | | 25 | 5 | | 26 | 36 | | 27 | 35 | | 28 | 7 | | 29 | 9 | | 30 | 20 | | 31 | 71 | | 32 | 25 | | 33 | 25 | | 34 | 14 | | 35 | 1 | | 36 | 38 | | 37 | 4 | | 38 | 16 | | 39 | 56 | | 40 | 46 | | 41 | 5 | | 42 | 34 | | 43 | 62 | | 44 | 41 | | 45 | 52 | | 46 | 42 | | 47 | 49 | | 48 | 57 | | 49 | 46 |
| |
| 100.00% | Passive voice overuse | Target: ≤2% passive sentences | | passiveCount | 3 | | totalSentences | 271 | | matches | | 0 | "been broken" | | 1 | "been caught" | | 2 | "been scraped" |
| |
| 100.00% | Past progressive (was/were + -ing) overuse | Target: ≤2% past progressive verbs | | pastProgressiveCount | 0 | | totalVerbs | 435 | | matches | (empty) | |
| 100.00% | Em-dash & semicolon overuse | Target: ≤2% sentences with em-dashes/semicolons | | emDashCount | 0 | | semicolonCount | 0 | | flaggedSentences | 0 | | totalSentences | 271 | | ratio | 0 | | matches | (empty) | |
| 94.00% | Purple prose (modifier overload) | Target: <4% adverbs, <2% -ly adverbs, no adj stacking | | wordCount | 2358 | | adjectiveStacks | 1 | | stackExamples | | 0 | "inside strange, viscous, holding" |
| | adverbCount | 53 | | adverbRatio | 0.022476675148430873 | | lyAdverbCount | 6 | | lyAdverbRatio | 0.002544529262086514 | |
| 100.00% | Repeated phrase echo | Target: ≤20% sentences with echoes (window: 2) | | totalSentences | 271 | | echoCount | 0 | | echoWords | (empty) | |
| 100.00% | Sentence length variance | Target: CV ≥0.4 for sentence word counts | | totalSentences | 271 | | mean | 9.07 | | std | 6.05 | | cv | 0.667 | | sampleLengths | | 0 | 10 | | 1 | 18 | | 2 | 4 | | 3 | 9 | | 4 | 28 | | 5 | 17 | | 6 | 12 | | 7 | 2 | | 8 | 3 | | 9 | 8 | | 10 | 10 | | 11 | 17 | | 12 | 2 | | 13 | 13 | | 14 | 8 | | 15 | 12 | | 16 | 5 | | 17 | 4 | | 18 | 22 | | 19 | 15 | | 20 | 9 | | 21 | 4 | | 22 | 10 | | 23 | 7 | | 24 | 3 | | 25 | 1 | | 26 | 9 | | 27 | 3 | | 28 | 4 | | 29 | 3 | | 30 | 7 | | 31 | 9 | | 32 | 5 | | 33 | 8 | | 34 | 3 | | 35 | 2 | | 36 | 19 | | 37 | 28 | | 38 | 9 | | 39 | 16 | | 40 | 2 | | 41 | 4 | | 42 | 4 | | 43 | 2 | | 44 | 9 | | 45 | 5 | | 46 | 8 | | 47 | 4 | | 48 | 7 | | 49 | 5 |
| |
| 42.99% | Sentence opener variety | Target: ≥60% unique sentence openers | | consecutiveRepeats | 19 | | diversityRatio | 0.2988929889298893 | | totalSentences | 271 | | uniqueOpeners | 81 | |
| 27.32% | Adverb-first sentence starts | Target: ≥3% sentences starting with an adverb | | adverbCount | 2 | | totalSentences | 244 | | matches | | 0 | "Then he bent back over" | | 1 | "Even the beads behind her" |
| | ratio | 0.008 | |
| 74.10% | Pronoun-first sentence starts | Target: ≤30% sentences starting with a pronoun | | pronounCount | 89 | | totalSentences | 244 | | matches | | 0 | "He broke left without warning" | | 1 | "She felt the worn leather" | | 2 | "She ignored it." | | 3 | "He took the footbridge at" | | 4 | "He hit the far side" | | 5 | "She loved the move and" | | 6 | "She snapped to it." | | 7 | "He feinted toward the main" | | 8 | "She went after him, boots" | | 9 | "He did not slow." | | 10 | "He rapped twice, quick, and" | | 11 | "He pressed it to the" | | 12 | "He slipped in." | | 13 | "It gave her nothing but" | | 14 | "She stepped back and took" | | 15 | "She adjusted the earpiece and" | | 16 | "She looked at the stairwell" | | 17 | "He blocked her as if" | | 18 | "She watched his hands and" | | 19 | "His weight sat easy on" |
| | ratio | 0.365 | |
| 50.16% | Subject-first sentence starts | Target: ≤72% sentences starting with a subject | | subjectCount | 200 | | totalSentences | 244 | | matches | | 0 | "He broke left without warning" | | 1 | "The satchel on his hip" | | 2 | "Harlow cut after him." | | 3 | "Spray hit her cheeks and" | | 4 | "Footsteps slapped, hers chasing his," | | 5 | "She felt the worn leather" | | 6 | "The strap had rubbed a" | | 7 | "She ignored it." | | 8 | "He took the footbridge at" | | 9 | "Cyclists swore, a dog jerked" | | 10 | "The satchel man looked back" | | 11 | "He hit the far side" | | 12 | "Harlow’s lungs ached, but the" | | 13 | "She loved the move and" | | 14 | "Every corner offered a line." | | 15 | "She snapped to it." | | 16 | "He feinted toward the main" | | 17 | "She went after him, boots" | | 18 | "A door sat ahead, steel," | | 19 | "He did not slow." |
| | ratio | 0.82 | |
| 61.48% | Subordinate conjunction sentence starts | Target: ≥2% sentences starting with a subordinating conjunction | | subConjCount | 3 | | totalSentences | 244 | | matches | | 0 | "Now a small black box" | | 1 | "If the Wardens clock you," | | 2 | "If they ask, you came" |
| | ratio | 0.012 | |
| 88.44% | Technical jargon density | Target: ≤6% sentences with technical-jargon patterns | | analyzedSentences | 105 | | technicalSentenceCount | 8 | | matches | | 0 | "He feinted toward the main road, then slid past a shuttered kebab shop and vanished down a concrete stairwell that swallowed sound." | | 1 | "He blocked her as if he had known her all his life." | | 2 | "A boy with a nose ring and a threadbare blazer, hare-bright eyes, the kind of brightness that had nothing to do with school." | | 3 | "That jerky energy, the twitch of a man who had spent weeks watching doors." | | 4 | "He gave her a look as if she had asked him to stop a tide with his fingers." | | 5 | "The black box in his hand looked older than the rats that made their lives between the rails." | | 6 | "He had long incisors, not sharp, just long, as if he had grown them for chewing a childhood too hard to swallow." | | 7 | "She tore free and drove forward into a space that smelled like the inside of a mausoleum just as the grate above slammed down and a wet voice close at her shoul…" |
| |
| 100.00% | Useless dialogue additions | Target: ≤5% dialogue tags with trailing filler fragments | | totalTags | 0 | | uselessAdditionCount | 0 | | matches | (empty) | |
| 100.00% | Dialogue tag variety (said vs. fancy) | Target: ≤10% fancy dialogue tags | |