Settings¶
Access settings from the gear icon in the navigation. Changes save immediately.
Providers¶
Storywright calls AI APIs directly from your browser — there's no Storywright server in the middle.
| Provider | What it is |
|---|---|
| NanoGPT (default) | 100+ models, one API key, pay-per-token |
| Google Gemini | Native integration — Gemini Pro, Flash |
| Anthropic | Native integration — Claude Sonnet, Opus, Haiku (with prompt caching) |
| OpenAI | GPT-4.1, o4-mini, embeddings |
| Ollama | Local models, free, offline-capable |
| LM Studio | Local models with GUI |
| Custom | Any OpenAI-compatible endpoint |
Add providers in Settings → Providers. Each gets a card where you enter the API key, base URL, and test the connection.
Web users: Local providers (Ollama, LM Studio) need CORS enabled. In LM Studio, enable the CORS toggle in the Developer tab. For Ollama, set
OLLAMA_ORIGINS=*before starting.
If you started with demo mode, the built-in provider is pre-configured. Add your own provider at any time — demo mode lifts as soon as you do, unlocking full model selection and removing any word limits.
See Choosing a Model for which models to pick.
Content Level¶
Controls what content is visible and what the AI can generate:
- 🌈 Kids — lighthearted themes only
- 📖 General — standard fiction (action, drama, romance)
- 🔓 Unrestricted — no restrictions
Content tags on stories, characters, and lorebooks are filtered against your level. Hidden content isn't deleted — just invisible until you change levels. Change in Settings → Content.
At the Unrestricted level, you can also create content override prompts — custom instructions that replace the default content boundaries.
Memory & Embeddings¶
Configure how the AI recalls facts from earlier scenes in Settings → Memory:
- Disabled (default) — keyword-only recall, no model needed
- In-Browser — runs a small model locally in your browser, no API calls
- API — uses your provider's embedding endpoint for best accuracy
If your provider doesn't support embeddings (e.g. Anthropic), the app automatically falls back to in-browser embeddings when available.
Other Settings¶
- Appearance — UI scale adjustment
- Writing Styles — browse and preview the available prompt styles (selected per-story, not globally)
- Story Structures — browse arc templates (Three-Act, Hero's Journey, etc.)
- Storage — browser local storage on web. Export your stories regularly — clearing browser data will erase them.
- Debug — file logging and verbose LLM request/response logging for troubleshooting
- About — version info, re-run the onboarding wizard, replay the guided tour