Anti-Zhuque Humanlike Writing and Self-Checks

Problem

Zhuque detection is sensitive to statistical fingerprints, sentence skeletons, and human feel. Mechanically replacing connectives or shortening sentences can still fail when text has even density, tidy structure, no mistakes, or polite dialogue. This skill rewrites how web-novel prose is expressed without changing the plot.

How It Works

• Human-like writing: adds rough speech, distraction, slow realization, fragmented dialogue, and flawed characters so voices feel specific.
• Hard self-checks: scans for em dashes, mechanical connectives, template phrases, empty summaries, parallelism, and “he knew / realized” conclusions.
• Punctuation discipline: dialogue must use 「」, with half-width "" removed and nested dialogue using 『』.
• Diagnosis tiers: uses density of banned terms, parallel passages, mental-state words, and dialogue share to choose repair strength, rewriting sections if needed.

Boundaries

It changes expression, not plot, character, or story direction. Best for Chinese web novels, character scenes, and dialogue-heavy chapters; not for formal essays, expository writing, or official documents. User hard rules such as chapter length, plain description, and no exposition must be applied alongside it.