Walks Chinese novelists from worldbuilding all the way to a finished book
Give it a genre, worldbuilding, and character setup, and it advances chapter by chapter through volume outlines, chapter outlines, prompts, manuscript, AI-trace removal, reader review, and archiving, delivering a Chinese novel manuscript that's ready to publish.
What this AI novel-writing workflow can do for you
awesome-novel is an AI writing workflow for Chinese novelists. It works the moment you open it — no software to install, no need to understand the technology behind it — helping you take a long-form novel from worldbuilding all the way to a finished book.
The most common problem with writing long-form fiction using AI is that once the conversation reaches several hundred thousand characters, the model can no longer remember your worldbuilding, character personalities, or the foreshadowing you've planted, and the characters drift as you keep writing. awesome-novel breaks this into 9 specialized agents: you start by saying "help me write a novel," and it confirms the genre, worldbuilding, main characters, and prose style with you, with 24 built-in genre profiles you can pick from directly; it then plans this volume's emotional arc, conflict ladder, and information gaps, before producing chapter outlines one by one, assembling prompts, writing the manuscript, and running AI-trace detection, optionally followed by a round of review from a "demanding reader," and archiving once it passes.
What you end up with is a plain-text novel project: story setup, volume outlines, chapter outlines, per-chapter prompts, and final manuscript, all in Markdown, ready to open and edit by hand. After each chapter is archived, it automatically updates character states, appends emotional arcs, and checks whether any foreshadowing is stale or due to be closed off, and the next chapter is still written according to the latest state.
It's suited to web-fiction authors, fan-fiction authors, and beginners just starting out who want to finish a long-form novel consistently with AI, and also to people who have already written half a book and want to import their draft and keep going. It isn't suited to just asking a question in a chat box and getting a paragraph, and it doesn't do grammar proofreading or literary criticism.
How it's different from just chatting with AI to write a novel
General-purpose AI chat can take on any topic, but its context is limited; once you're dozens of chapters in, it struggles to consistently remember character personalities, timelines, and the foreshadowing you've planted. awesome-novel turns all of this into project files that travel with the book, automatically appending character states and emotional arcs after each chapter is archived; if you stop midway, you can restart and continue from the recorded progress without having to retell the story so far.
Most other AI novel-writing tools or prompt templates simply hand the model a one-off prompt. This workflow's 9 agents each have their own role, and it includes author confirmation checkpoints: after the setup, volume outline, and chapter outline are written, it pauses to show you a summary, and only continues once you approve it — if something's wrong, you just say so and it revises; only after confirmation does the pipeline of prompts, manuscript, AI-trace removal, and archiving run in sequence. When you want to save effort, say "you write it all" and it handles the confirmations for you.
Text generated by general-purpose chat often carries a mechanical tone and formulaic phrasing. awesome-novel has built-in anti-AI-trace rules, and after writing there's a separate stage that runs multiple checks and quantitative scoring; it also records how you edit the AI's original text and distills that into your own personal avoidance rules, with high-frequency ones settling into long-term memory.
If you want to seriously finish a long-form novel and are willing to let AI write in your style and at your pace, choose this. If you only occasionally generate a bit of copy or practice your writing, general-purpose chat is faster.
Common questions about writing Chinese novels with AI
9 questions
What is awesome-novel?
It's a writing workflow for collaborating with AI on long-form Chinese novels; it works the moment you open it, with nothing to install or configure. It divides setup discussion, volume outlines, chapter outlines, prompts, manuscript writing, AI-trace detection, and archiving among 9 specialized agents, rather than generating everything at once in a single chat box.
Do I need to know how to code to write a novel with awesome-novel?
No. You can start writing the moment you open it, with no tools to install and no runtime environment to configure. You just tell it the genre and rough idea you have in mind, and awesome-novel confirms the worldbuilding, characters, and prose style with you before advancing chapter by chapter.
Which genres does it support?
It has 24 built-in genre profiles covering mainstream categories such as xianxia, urban, mystery, historical, sci-fi apocalypse, and Western fantasy, each including character archetype tendencies, narrative tone, and chapter prompt templates. You can also skip the profiles and simply describe the genre you want to write, and awesome-novel will configure a writing style and pacing template for it.
I've already written half a book — can I import it and continue?
Yes. Put your existing draft into a novel project and say "import this novel," and awesome-novel will continue writing from there. If the project is in the older 2.x format, on first open after upgrading it will automatically detect this and guide the migration to the 4.0 format, backing up the old files first and copying the archived manuscript over as is.
Will it change the worldbuilding and characters I've settled on without asking?
No. For the three stages that determine where the story goes — setup, volume outline, and chapter outline — awesome-novel must pause after writing and show you a summary, and only moves to the next step once you reply "okay"; if you want changes, it reschedules the relevant stage, revises, and confirms again. It also won't let the orchestrating novel director write the setup directly; any changes are discussed with you first.
Does the Chinese it generates sound AI-written?
It defaults to a low AI-trace feel. Anti-AI rules are injected when prompts are assembled, the writer self-checks while drafting the manuscript, and after writing there's a separate stage that runs multiple checks and quantitative scoring. It also records how you edit the AI's original text each time and distills that into your own personal avoidance rules, with the most frequent ones settling into long-term memory.
Can it write in my own prose style?
Yes. Put the sample writing you want it to learn from into the samples directory of the novel project, and awesome-novel will distill quantitative style parameters — such as sentence length, dialogue ratio, and adjective density — write them into a style master card, and inject them into prompts by scene. After archiving, it also calibrates incrementally using your final drafts, getting closer to your way of expressing things the more it writes.
When writing a long novel, does it still remember earlier characters and foreshadowing later on?
Yes. The setup, character states, timeline, and foreshadowing ledger are all stored as plain-text files in the novel project; after each chapter is archived, it automatically appends character state history and emotional arcs, and scans for unfulfilled and newly planted hooks. These files also serve as checkpoints — if you stop midway and restart, it continues from the recorded progress without you having to recap the story so far.
I just want to write a short piece — does it support that too?
Yes — short-form and long-form are two separate workflows that never mix. Say "write a short story" or "write a short piece," and awesome-novel creates a separate short-form project; its flow is: set the mood, brainstorm and outline, assemble the contract, audit the contract, write in batches, polish out AI traces, check the payoff, and accept delivery — and the set of agents it uses is not shared with the long-form one.