Intelligent Novel Assistant

Problem

Web novel writing often stalls on platform rules, pacing, and word counts: Tomato, Qimao, Qidian, JJWXC, and Zhihu expect different chapter lengths and opening behavior. Long drafts can also lose track of character arcs, conflict, and hooks. Intelligent Novel Assistant moves those constraints into the workflow instead of relying on late rewrites.

How It Works

• Platform first: confirm the publishing platform, then read platform-profiles.md to set chapter length, pacing, and opening expectations.
• Seven questions: lock in platform, emotion label, genre, protagonist, contrast, core conflict, and chapter count, with mismatch warnings.
• Planning artifacts: produce 00-大纲.md, 01-人物档案.md, 02-情绪曲线.md, and 03-黄金开篇.md as a reviewable scaffold.
• Chapter loop: run analyze, draft, polish, self-check, and close-out steps, with early 20% conflict, ending hooks, concrete dialogue, and less generic prose.
• Word-count checks: use scripts/check_chapter_wordcount.py before delivery.

Boundaries

It suits structured serial web-novel workflows, but it does not fetch live platform metrics. Data feedback every 5 chapters must be supplied by the user. For non-serial fiction or fully freeform writing, the platform constraints may be too prescriptive.