Use agents to keep a long novel organized

An author used Claude Code to plan, draft, and maintain continuity for a 700,000-character apocalypse novel.

Done with Claude Code· Advanced· Oct 2026

What they did
The author began by having Claude Code examine Fanqie novel rankings, study books from the same genre, and read TV Tropes, then used Socratic questioning to build the outline. The project was run from markdown files, with Claude Code updating the timeline, character records, and foreshadowing notes as the manuscript grew. For each chapter, the process moved through planning, a plan review, a 2,500-to-3,500-character draft, checks from editor, setting, and reader agents, a targeted rewrite, and then wiki updates.
How it went
Over one holiday, the author completed a 700,000-character apocalypse novel. They said the file structure and revision loop kept the work from collapsing, but the source did not mention reader feedback or quality results. A cleanup pass was still needed because the first version sounded too machine-written, and broad instructions such as “make it more exciting” were not useful.
Tips
A shared file system makes it easier to preserve continuity in a long project. Use specific revision requests, and stop for approval before any action that would overwrite or remove content.

Try it yourself with Claude Code

I’m working on a [genre] novel of around [length] characters. In this folder, help me set up and keep current markdown files for the timeline, character list, and foreshadowing tracker, and then handle each chapter I give you as a cycle of planning, drafting, editing, and setting review. Start by asking me questions that will turn [chapter outline] into a clear plan, write that chapter at [target length] characters, refresh the tracking files, and show me the proposed changes before replacing anything already saved. If a step would send, delete, submit, or overwrite content, stop and ask for my approval first.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on cloud.tencent.com ↗

1 more report of this

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.