Getting Started with Storywright¶

Welcome to Storywright — an AI-powered fiction writing workbench that helps you plan, generate, and revise stories scene by scene.

What is Storywright?¶

Storywright is a cross-platform app for writing fiction with AI. You build a story by defining characters, world details, and a premise, then let an AI generate each scene with full context — including character cards, lorebook entries triggered by keywords, extracted memories from earlier scenes, and continuity tracking for every entity.

Here's what makes it different:

Scene-by-scene generation — you plan the story structure first, then generate one scene at a time with the AI aware of everything that came before.
Characters and world-building — create rich character cards and lorebooks (keyword-triggered world info) that automatically feed into the AI's context.
Memory and continuity — the app extracts facts from generated text and tracks structured state (location, appearance, possessions, relationships) so your story stays consistent across scenes.
Works with any OpenAI-compatible API — bring your own API key. The default provider is nano-gpt, but any compatible endpoint works.
Cross-platform — runs on macOS, iOS, Windows, Android, and web from a single Flutter codebase.
No backend server — the app calls LLM APIs directly from your device. Your data stays local.

Installing and Launching¶

Download¶

Storywright is available for all major platforms:

macOS — download the .app bundle or build from source
iOS — install from TestFlight or build from source
Windows — download the installer or build from source
Android — install the APK or build from source
Web — access via the hosted web app or build and serve locally

Build from Source¶

Storywright is a Flutter app. If you have the Flutter SDK installed:

Clone the repository
Run flutter pub get to install dependencies
Run flutter run -d macos (or ios, windows, android, chrome)

First Launch — Onboarding Wizard¶

The first time you open Storywright, a short onboarding wizard walks you through setup:

Welcome — an overview of Storywright and the writing workflow.
API Setup — enter your API key and base URL. The default endpoint is https://nano-gpt.com/api/v1. You can test the connection before continuing.
Content Level — choose your content level (Kids / General / Mature / Unrestricted). This step may be skipped if the content level is locked by build config.

After completing the wizard, a Guided Tour launches automatically. The tour walks you through the workspace using interactive highlights — showing you where to find stories, characters, lorebooks, and generation controls.

You can skip the wizard at any point. To re-run it later, go to Settings → About → Re-run Setup Wizard. To replay the guided tour, use Settings → About → Take a Tour.

The app has three main tabs:

1. Stories (default)¶

This is where you spend most of your time. The left side shows your story list; the right side is the story workspace where you plan, generate, and revise scenes.

2. Library¶

Your reusable content, organized into two sub-tabs:

Characters — your global character card library. Characters created here can be added to any story.
World Info — all lorebook entries across all worlds and characters, with source badges so you can see where each entry comes from.

3. Settings¶

App configuration, organized into four groups:

Connection — Providers, Models
Creative — Prompts, Structures, Content
System — Appearance, Storage, Memory, Debug
Info — About

macOS — native sidebar navigation
Tablet / Desktop — navigation rail
Mobile — bottom navigation bar

Your First Story — Walkthrough¶

Follow these steps to create a complete story from scratch. The whole process takes about 15 minutes.

1. Create a story¶

Click + New Story in the Stories sidebar. Give your story a name — something descriptive like "Murder at Crestwood Harbor." A default world is assigned automatically.

2. Set the premise¶

In the left panel, write a premise describing what your story is about. This is the seed the AI uses for planning and generation.

Example: "A detective investigates a murder in a small coastal town where everyone has something to hide."

3. Add creative direction¶

Optionally, add style hints that guide the AI's approach. These shape the tone and structure without constraining the plot.

Example: "Noir atmosphere, sharp dialogue, twist ending"

4. Choose a writing style¶

Pick a prompt style that matches the kind of story you want to write:

Literary fiction — rich prose, thematic depth
Genre fiction — plot-driven, clear stakes
Sci-fi — ideas-driven science fiction, worldbuilding rigor
Mystery-thriller — propulsive mystery, fair-play clues, escalating stakes
Fantasy — immersive fantasy, magic with rules, deep worldbuilding
Humor — witty, comedic timing
Romance — emotional beats, relationship focus
Horror — tension, dread, atmosphere
Minimalist — spare, precise language
Audio storytelling — optimized for spoken delivery

5. Add characters¶

Click + Add Character in the Character Panel. You can:

Create a new character — fill in personality, description, and backstory right here.
Pick from your library — reuse a character you've already created.

Assign each character a role: protagonist, antagonist, supporting, mentor, or any custom role. Roles help the AI understand each character's function in the story.

6. Plan the story¶

Click the Plan button in the toolbar. The AI generates a scene-by-scene outline based on your premise, characters, and creative direction. Each scene gets a title, a brief outline, and a target word count.

7. Review and edit the plan¶

Before generating, fine-tune the plan:

Reorder scenes by dragging them into a new sequence
Edit outlines to adjust what happens in each scene
Change word counts to make scenes longer or shorter
Add or remove scenes to reshape the story structure

8. Generate¶

Click Generate to write the first scene. Text streams in real-time as the AI writes. Once generation finishes, the app automatically extracts:

A summary of the scene
Memories — key facts for future recall
Continuity state — updated locations, appearances, possessions, and relationships for each entity

9. Read and revise¶

Read through the generated text. You have two ways to revise:

Revision box — type instructions like "add more tension to the opening" or "rewrite the dialogue between Sarah and Marcus" and the AI will revise accordingly.
Direct editing — edit the text yourself, just like in any text editor.

10. Continue¶

Move to the next scene and generate again. Each scene builds on the full context from previous scenes — summaries, memories, continuity state, and active lorebook entries all carry forward automatically.

11. Export¶

When your story is complete, use the export menu to save it as Markdown or plain text.

Key Concepts¶

Concept	What it means
Story	A collection of scenes with a premise, characters, and a world.
Scene	One chapter or section of your story, generated individually with full context from prior scenes.
Scene Plan	An outline entry for a scene — its title, what happens, and a target word count.
Character Card	A character's personality, description, and backstory. Included in the AI's context during generation.
Lorebook	A collection of world-info entries triggered by keywords. When a keyword appears in the current scene context, that lorebook entry is automatically injected into the AI's prompt.
Memory	Facts extracted from generated scenes. Recalled by semantic similarity when generating future scenes to maintain story coherence.
Continuity	Structured state tracking — location, appearance, possessions, relationships, and more — for each entity across scenes. Prevents contradictions.
Writing Style / Prompt	The system instructions that shape the AI's writing tone, structure, and approach.

Next Steps¶

Now that you have the basics, dive deeper into each area:

Stories Guide — detailed story workflow, scene management, and generation options
Characters Guide — creating, importing, and managing character cards
Lorebooks Guide — world-building with keyword-triggered entries
Settings Guide — API configuration, model selection, and preferences