# How to Vibe Code Your Own Novelcrafter (and Stop Paying for It)

> Your Novel Writing Software

- Site: https://novelcrafter.com
- Category: Writing & Productivity SaaS
- Verdict: **Solid side project** (68/100 vibecodeable)
- Estimated effort: 2-3 weeks part-time

## Verdict

You can build a robust personal clone of Novelcrafter with an AI coding agent, though managing the real-time Codex mention parser and dynamic RAG context injection will test your debugging patience.

Novelcrafter is a feature-rich workspace that combines a custom rich text editor, a structured wiki database (The Codex), visual planning grids, and a multi-model LLM chat integration. An AI coding agent can scaffold the Next.js frontend, Turso database schema, and LLM proxy routes cleanly. However, getting the background text-parsing engine to cross-reference manuscript text against Codex entity aliases on the fly without lagging keystrokes, and constructing the dynamic RAG context payloads for the Workshop chat, will require iterative prompt engineering and state management fixes across multiple weekends.

### What you can't replicate

- The 200k-author community and specialized prompt libraries
- Years of edge-case polish across hundreds of obscure LLM model variations
- Pre-existing series bibles and user writing histories

## What it does

An AI-assisted web-based writing platform combining a distraction-free rich-text manuscript editor with an intelligent story bible (The Codex) that automatically tracks characters, locations, lore, and relationships across a series.

### Core features

- Distraction-free rich text manuscript editor with marker/highlighter support
- The Codex: automated intelligent wiki tracking characters, locations, and items
- Dynamic progressions and timeline tracking for Codex entries across a series
- Planning dashboards (Grid drag-and-drop board and Matrix spreadsheet view)
- Multi-vendor LLM integration layer with Bring-Your-Own-Key (BYOK) and prompt templates
- Workshop Chat panel for contextual brainstorming against manuscript and Codex data
- Prose review tools for detecting overused words, dialogue tags, and AI-isms
- Project import and export handling Markdown, HTML, and Word formats

## The business

### Pricing

- Scribe: $4/mo — Unlimited books, series & universes, Codex, basic review features, collaboration viewing/editing.
- Hobbyist: $8/mo — Everything in Scribe plus AI integration (BYOK), scene summarization, character extraction, and generation from scene beats.
- Artisan: $14/mo — Everything in Hobbyist plus interactive Workshop Chat features and advanced review features.
- Specialist: $20/mo — Everything in Artisan plus collaborative writing capabilities and team management.

### Funding

$0 raised.

Founded 2023.
Team size: ~8.

## The hard parts

- RAG context assembly: dynamically chunking and injecting series bibles, outline notes, and scene summaries into LLM payloads without breaking context windows
- Real-time manuscript parsing: scanning typing input on the fly to detect entity mentions and link them to Codex databases without UI lag
- Multi-vendor LLM proxy orchestration supporting streaming responses, rate limits, custom error states, and local endpoints like Ollama
- Multi-pane workspace state management uniting rich text editing, grid boards, matrix spreadsheets, and side chat panels

## How to vibe code Novelcrafter

### Prerequisites

- Node.js (free): Required runtime for executing the Next.js full-stack development environment.
- GitHub (free): Hosts the git repository and connects to deployment pipelines.
- OpenAI / Anthropic API Key (pay-as-you-go): Provides language model access for the Workshop chat and scene summarization features.

### Recommended AI tools

- Claude Code: Best-in-class agentic coding tool for scaffolding multi-file schema architectures, rich text editor components, and backend proxy routes.
- Cursor: Provides fluid inline diff reviews and iterative tweaking for complex multi-pane React workspace layouts.

### Stack

- Frontend: Next.js with Tailwind CSS and TipTap rich-text editor components
- Backend: Next.js API Routes / Server Actions
- Database: Turso (libSQL/SQLite at the edge for per-user local-first relational speed)
- Auth: better-auth
- Payments: none (personal use clone)
- Other: Vercel AI SDK for multi-provider streaming LLM integration, Zod for schema validation

### Hosting

- Cloudflare (Hosting the Next.js application frontend and Turso database backend on a fast edge network.): $0/mo

### Build guide

1. **Project Scaffolding and Database Schema** — Initialize the Next.js project with Tailwind CSS, configure Turso database connectivity, and establish the core relational schema for users, books, chapters, scenes, and Codex entries.

```
Scaffold a new Next.js 16 application using App Router, TypeScript, and Tailwind CSS v4. Set up a Drizzle ORM schema connected to Turso (libSQL). Create the database models for users, books (with fields for title, description, cover image, target word count), chapters (ordered sequence within books), scenes (ordered sequence within chapters, containing rich text content blobs, status, POV character, word count), and codex_entries (categories like characters, locations, lore; custom metadata key-value fields; aliases array; progressions timeline). Include proper foreign key constraints, indexing on book_id and chapter_id, and migration scripts. Ensure all TypeScript types are fully exported.
```

2. **Authentication and Dashboard Shell** — Implement authentication using better-auth and build the main multi-pane dashboard layout featuring book management, sidebar navigation, and project switching.

```
Implement email/password authentication using better-auth integrated into the Next.js app. Create secure API routes and client-side session providers. Build a responsive dashboard layout with a collapsible sidebar for managing books and series universes. Include a main bookshelf grid view displaying book covers, current word counts, progress bars against goals, and a 'New Book' modal dialog. Ensure clean routing and protected layout guards for authenticated routes.
```

3. **Rich Text Manuscript Editor & Planning Views** — Build the distraction-free rich-text writing canvas using TipTap, alongside the Grid and Matrix planning boards for structuring acts, chapters, and scenes.

```
Build a distraction-free manuscript writing interface using a TipTap rich-text editor component with toolbar options for bold, italic, headings, quotes, and custom highlights/markers. Implement debounced auto-saving to update scene content and calculate live word counts in the database. Add split/pin panel capabilities allowing users to view notes side-by-side with the text. In addition, create a 'Planning' view with a drag-and-drop Grid board for reordering chapters and scenes, and a Matrix spreadsheet view mapping characters against scenes for high-level structural overview.
```

4. **The Codex Wiki & Automatic Mention Tracking** — Develop the Codex story bible interface and a text parser that detects entity mentions in the manuscript and links them interactively.

```
Build the Codex story bible manager supporting custom categories (Characters, Locations, Lore, Objects), rich-text backstories, quick facts, tags, thumbnail images, and a timeline progressions sub-system for tracking changes across a series. Implement a background text-scanning utility that parses manuscript content against Codex entry titles and aliases, returning matching positions so the frontend can highlight entity mentions interactively and provide one-click navigation from manuscript text directly to the corresponding Codex entry.
```

5. **Multi-Vendor LLM Integration & Workshop Chat** — Set up the Vercel AI SDK abstraction layer to support custom API keys (OpenAI, Anthropic, OpenRouter, Ollama) and build the contextual Workshop Chat panel.

```
Implement a multi-vendor LLM connection service using the Vercel AI SDK supporting OpenAI, Anthropic, OpenRouter, and local Ollama endpoints via user-supplied API keys stored securely. Build a 'Workshop Chat' slide-over panel that allows users to converse with an AI assistant. Implement automated RAG context assembly where the backend injects the current scene content, book outline, and active Codex entries into the system prompt context window. Support streaming chat responses, custom prompt template management, and one-click extraction of generated scene beats or Codex entries back into the project database.
```

### Cost vs paying

**Starting costs (one-time):**

- Domain name (optional): $12/yr
- AI coding tool subscription: $20 one-time
- Total: ~$32 one-time

**Ongoing costs (monthly):**

- Cloudflare & Turso free tiers: $0/mo
- LLM API usage (BYOK/OpenAI/Anthropic): ~$5-15/mo depending on writing volume
- Total: ~$10/mo

- Paying for the SaaS instead: $4/mo to $20/mo
- Build time: 25-35 hours
- AI tool credits: $20 (Claude Pro or Cursor Pro)
- Break-even: N/A (Built for personal learning and custom workflow control)

## Sources

- [Novelcrafter Home Page](https://novelcrafter.com)
- [Novelcrafter Pricing Page](https://novelcrafter.com/pricing)
- [Novelcrafter Features Page](https://novelcrafter.com/features)