novel-writer

Original🇨🇳 Chinese
Translated
3 scriptsChecked / no sensitive code detected

An intelligent assistant designed for long-form online novel creation, supporting full-process management from setting generation to main text writing, including intelligent quality control and state synchronization. (Yunshu Optimized)

9installs
Added on

NPX Install

npx skill4agent add cyangzhou/-2--project-yunshu- novel-writer

SKILL.md Content (Chinese)

View Translation Comparison →

Novel Writer AI Novel Creation Execution Process (Optimized Version)

Core Objective

Clarify the division of labor between users and AI, standardize the operation steps, input and output requirements, and document formats of the entire creation process, ensure that AI accurately performs tasks such as creation, revision, and quality control, and guarantee creation consistency and efficiency.

I. Core User Operations: Select Creation Mode

Mode 1: Create a New Book (Complete Initialization Process)

1. User Input Submit basic creation information:
  • Book title
  • Protagonist name
  • Full story outline
  • Core character list (with core traits marked)
2. AI Execution Steps
  • Step 1: Initial Setting Generation
    • Content: Worldview settings, full-book plot outline, protagonist growth journey, supporting character growth journey (including side plots + clear appearance chapters).
    • Format: Markdown document
    • Naming:
      【Book Title】- Initial Settings.md
  • Step 2: Setting Revision (Loop)
    • Trigger: User provides revision suggestions for the initial settings.
    • Content: Correct errors based on suggestions and re-output the complete content.
    • Naming:
      【Book Title】- Initial Settings.md
      (overwrite and update)
  • Step 3: Volume-by-Volume Creation Preparation
    • Trigger: User replies "Confirm" to the final version of the initial settings.
    • Prerequisite: Read the [Final Version Initial Settings].
    • Content:
      • Detailed chapter outline for the current volume (covering the entire volume)
      • Character profile for the current volume (characters appearing in this volume)
      • Character relationships (dynamically updated)
      • Original Work Deviation Guide (for fanfiction only)
    • Naming:
      【Book Title】- Volume X - Creation Guide.md
  • Step 4: Volume Preparation Revision (Loop)
    • Trigger: User provides revision suggestions.
    • Content: Revise based on suggestions and re-output the complete content.
    • Naming:
      【Book Title】- Volume X - Creation Guide.md
      (overwrite and update)
  • Step 5: Main Text Generation
    • Trigger: User replies "Confirm" to the final version of the volume guide.
    • Prerequisite: Read the [Final Version Volume Creation Guide].
    • Content: Main text of the corresponding chapters in the current volume (each chapter ≥ 4000 words, consistent with setting logic).
    • Format: TXT document
    • Naming:
      【Book Title】- Volume X - Chapter X - [Chapter Title].txt

Mode 2: Modify an Existing Book

1. Interaction Process
  • User Operation: Select a book -> Select a module (outline/character settings/worldview, etc.) -> Manually modify content and save.
  • AI Inquiry: "Do you need AI optimization for the selected module?"
2. AI Execution Steps
  • If user selects "Yes":
    • Automatically optimize the module content, maintain consistent core logic, and improve expression quality.
    • Output: Markdown document (Naming:
      【Modified Module Name】.md
      )
  • If user selects "No":
    • Only retain the user's modified content without additional processing.

Mode 3: Add/Modify Main Text

1. Interaction Process
  • User Operation: Select a book -> Select a chapter (modify existing or add a new chapter).
2. AI Execution Steps
  • Prerequisite Reading: Automatically read the book's
    Detailed Chapter Outline
    ,
    Current Character Profile
    ,
    Character Relationships
    , and
    Original Work Deviation Guide
    .
  • Task Execution:
    • Modify Main Text: Rewrite/polish based on user suggestions, each chapter ≥ 4000 words.
    • Add Main Text: Automatically generate a new chapter based on the detailed outline, chapter number increments by 1, each chapter ≥ 4000 words.
  • Output: TXT document
  • Naming:
    【Book Title】- Volume X - Chapter X - [Chapter Title].txt

II. AI Automatic Execution: Intelligent Quality Control Process

Trigger Mechanism: After AI completes any [Setting Generation], [Volume Preparation], or [Main Text Generation], automatically initiate the following process.

1. Archivist Check (Consistency Check)

  • Task: Compare the output content with the original/final settings.
  • Detection Items: Character setting deviations, plot logic conflicts, inconsistent settings.
  • Output:
    • Markdown document: Add a
      ## Quality Control Notes
      section at the end.
    • TXT document: Add a short
      [Quality Control Remarks]
      at the end.

2. Quality Audit

  • Task: Optimize content quality (only for main text).
  • Optimization Items:
    • Emotional intensity (adapt to the plot)
    • Hook design (enhance anticipation)
    • Narrative rhythm (fast-slow transition)
  • Restriction: Only optimize expression and rhythm, strictly prohibited to alter core setting information.

3. State Sync

  • Task: After the main text is completed, extract new character details (personality/experiences) and plot changes.
  • Write-back: Update to the corresponding
    Character Profile
    and
    Character Relationships
    documents.
  • Output:
    • Generate/continue the log:
      【Book Title】- State Sync Log - Volume X Chapter X.md
    • Record the specific synchronized content and modification positions.

III. Key Execution Instructions

  1. Document Naming Specification: Strictly follow the format
    【Book Title】- Document Type/Volume & Chapter Information
    .
  2. Revision Feedback Requirements: User feedback must include
    Target Module
    +
    Specific Revision Direction
    .
  3. Exception Handling: When encountering ambiguous requirements or setting conflicts, must actively ask for clarification, and it is strictly prohibited to make assumptions without authorization.