Docs
Documentation
Settings and backup
Use Settings to adjust appearance, editor behavior, language, general preferences, and backup.
Appearance
Choose light, dark, or automatic mode and adjust the visual theme of the application.
Light, dark, or automatic: choose whether the application follows the system or uses a fixed mode.
Preset themes: select from built-in visual palettes.
Customization: adjust primary color, border intensity, backgrounds, and surfaces.
Import and export theme: save a visual setup or reuse an existing one.

Editor and preferences
Editor typography: adjust font, text size, spacing, and reading width.
Editor behavior: configure autosave, spellcheck, smart punctuation, toolbar mode, and related preferences.
General preferences: language, color-blind mode, outline density, notifications, and panel behavior.
AI
The AI tab is optional. It centralizes provider, model, credentials, and connection testing for people who want assisted research, Ask mode, or Book Chat memory.
Where to configure it: open
Settings → AIto centralize provider, model, key, timeout, and connection testing in one place. If you configure nothing, the app still works for writing, organization, metrics, and export.Provider and model: choose between Anthropic, OpenAI, Gemini, or Ollama. With cloud providers, the required calls go directly to the service you choose; Inkthar does not run its own AI server between your writing and the provider.
Your API key: after selecting a cloud provider, paste the key from your own account, adjust timeout if needed, and save. The app indicates when a key is already configured without exposing the full value. Review the chosen provider's privacy, retention, and model-training controls when available.
Local Ollama: if you prefer to avoid cloud AI providers, provide the local instance URL, list the available models, and select which one to use. In this case, no API key is required and model responses run on your own machine/local environment.
Test before using: after saving, use the built-in test to validate the setup before relying on it in assisted research, contextual search, or Ask mode.
Where this is used: the chosen configuration powers assisted research, contextual search, Ask mode, and Book Chat memory. It does not enable automatic writing, chapter rewriting, or ready-made passage suggestions.
Book Chat memory: to find passages before answering, the app uses embeddings. OpenAI, Gemini, and Ollama can generate embeddings; with cloud providers, the required passages are sent to the provider you choose.
Claude and embeddings: Anthropic can be the main answer provider, but it does not provide embeddings in the app. If you use Claude with Book Chat, configure a separate memory source: OpenAI, Gemini, or Ollama.
Contextual RAG: when enabled, it adds narrative context to indexed passages and uses extra AI calls while preparing memory.
What can leave the computer when using AI
| Feature | Data used outside the local app | How to control it |
|---|---|---|
| Assisted research and Ask mode | Query, typed context, categories, and data required to generate or refine the answer; web search also queries external sources. | Use only when needed, choose the provider, and review its policies. Without AI configured, these flows do not run. |
| Book Chat | Question, recent history, current selection when available, project metadata, and relevant passages retrieved from memory. | Use Ollama to keep processing in your local environment, or a cloud provider under your own account. |
| Memory embeddings | Manuscript passages used to generate search vectors when the embedding provider is OpenAI or Gemini. | Choose Ollama as the embedding provider if you want to avoid this cloud step. |
| Contextual RAG | Passages being prepared can receive additional context through extra calls to the configured LLM. | Keep disabled if you want fewer AI calls during indexing. |
| Local Ollama | The app talks to the configured local URL for model and embedding calls. | Depends on your Ollama environment; no cloud-provider key required. |
License
The License tab shows the current installation status and centralizes activation, validity, and machine binding.
Current status: see whether the installation is in trial, active, expired, or deactivated state, including dates and license origin.
Activation and validity: the area shows partial key, activation date, last validation, and expiration when applicable.
Machine binding: you can also review the device identifier and deactivate this machine before moving to another environment.
Backup and maintenance
This area covers backup export, data restore, update checks, and preference reset.
Export backup: generate a <code>.zip</code> file with your local data so you can keep a complete copy of the work outside the app.
Restore backup: when you need to return to an earlier state, you can import a saved backup and bring the project back into Inkthar.
Backup contents: the file includes the SQLite database, covers, entity images, and research files stored under
storage.Restore safety: restoring replaces the current local state with the imported backup. The app validates the file and attempts to preserve the previous state if restore fails, but exporting a backup before restoring is still recommended.
Updates: the app also keeps update-check preferences here so installation maintenance stays in the same panel.
Reset preferences: if you feel you experimented too much, you can return settings to their defaults without erasing the book's content.
