Web Novel Writing Engine

Problem

Writing web novels often stalls at concrete steps: writers may not know which opening, pacing, or payoff patterns perform well on a target platform, may have only a rough premise, may produce chapters that trip platform review, or may get prose that sounds formulaic.

How It Works

This skill turns the workflow into explicit stages. It can analyze top-ranked novels for shared patterns in opening structure, chapter rhythm, character setup, and payoff distribution; then check an outline for main plot, subplots, early hooks, motivation, and world consistency, expanding it by volume, chapter, or scene. For drafting, it generates chapters around 1500-2000 words, balancing dialogue, pacing, and end-of-chapter hooks, with Jjwxc used as a high-strictness reference. After drafting, it runs layered checks for banned themes, tag-content consistency, platform-specific expectations, and rewrites violating wording instead of deleting plot. The de-AI pass scans for template sentences, generic emotion lines, over-dense description, and missing colloquial texture, replacing them with concrete actions, restraint, and small human details. It can also create Mermaid or Tencent Docs mind maps for character relations, timelines, or worldbuilding.

Boundaries

It is suited to chapter-by-chapter drafting, staged review, and pre-release proofing, not to bulk-generating hundreds of chapters or copying specific works. Ranking analysis depends on agent-browser access to public leaderboards, and can fall back to user-supplied titles if scraping fails. Authors remain responsible for final publication, and platform rules may change.