AI-Novel-Writing-Assistant: a director-style AI production system for managing the full lifecycle of long-form novel writing

What it solves

This project addresses the difficulty of writing long-form novels with AI, moving beyond simple "prompt-and-response" chat interfaces that often lead to inconsistent plots and fragmented narratives. It provides a structured, "director-style" production system designed to help beginners complete an entire book by managing high-level planning, world-building, and character consistency across hundreds of chapters.

How it works

The system uses an AI-native workflow orchestrated by LangGraph and LangChain to transform a single spark of inspiration into a full novel. It operates through a series of interconnected stages:

• Automatic Director: Generates the overall direction, world-building, and chapter-level tasks from a simple idea.
• Production Pipeline: A chain of generation, auditing, and fixing that can be paused and resumed at checkpoints.
• Creative Hub: A central agent runtime that routes natural language intents to specific planning or execution tasks.
• Long-term Assets: Uses RAG (via Qdrant) and a "Style Engine" to maintain a knowledge base of world settings, character archives, and specific writing styles that are recalled during generation to ensure consistency.
• Feedback Loop: Every completed chapter feeds back character states, facts, and plot hooks into the global context to influence subsequent chapters.

Who it’s for

• Novice Writers: People who have ideas but lack the experience to structure a long-form novel.
• AI Developers: Those interested in implementing complex Agent Workflows, LangGraph orchestration, and long-chain task management in a real-world application.

Highlights

• End-to-End Production: Covers everything from initial inspiration to macro-planning, volume strategy, chapter execution, and quality repair.
• Character Evolution Tracking: Scans chapters to track changes in character appearance and status, generating consistent visual references.
• Style Engine: Allows users to extract and save writing style features from existing text to avoid generic "AI-sounding" prose.
• Crossover Workflows: Includes workshops for adapting completed novels into manga or short drama scripts.
• Multi-Model Routing: Supports routing different tasks (planning vs. writing vs. auditing) to different LLM providers (OpenAI, DeepSeek, etc.).