Image keys make pictures. A text key makes the words, the page plans and the metadata. They are configured separately on the Providers tab.
What a text key unlocks:
- The AI story writer, which drafts the whole story, splits it into pages, and returns a separate visual scene description for each illustration
- AI planned scene lists for coloring books, so the pages tell a coherent visual sequence instead of repeating
- AI planned activity pages, which choose the puzzle type and write the kid facing instruction for each page
- The KDP listing kit on the Publish tab
- The AI Rewrite button on individual page text boxes
The three routes:
- Anthropic, Claude. Saved in its own row on the Providers tab.
- OpenAI, GPT. Saved in its own row on the Providers tab.
- Google, Gemini. This one has no separate row, because the text engine reuses the Google key you saved under Image Providers. One free Google AI Studio key therefore powers both the character locked illustrations and every text feature.
The plugin tries Anthropic first, then OpenAI, then Google, and uses the first one that has a key saved. Any single key is enough.
What happens without a text key?
Nothing breaks. The text features simply stay unavailable and the plugin falls back where it can. Coloring books derive their scene list locally so they still generate with zero keys, activity books fall back to a local page plan, and the listing kit and AI Rewrite return a clear message telling you to add a key. Story books do require a text key, because there is no local fallback for writing a story.