What is a Storybook?
A Storybook is a knowledge base attached to your creation, made up of multiple independently configurable entries. During a conversation, entries are automatically injected into the AI's context based on trigger conditions, supplementing background knowledge, world-building, or scene information.
Entries
A Storybook is composed of multiple entries. Each entry is a standalone piece of knowledge that can be individually configured with its own trigger behavior and injection settings.
Core Fields
| Field | Description |
|---|---|
| Name | The entry's identifier label, for management purposes |
| Content | The text actually injected into the AI's context |
| Trigger Mode | Constant (always inject) or Keyword-triggered |
| Trigger Keywords | Only active in Keyword mode — injection fires when matched |
| Trigger Probability | 0–100%, controls the actual probability of injection when triggered |
| Insertion Position | Where in the context the entry content is injected |
| Message Role | Injected as system / user / assistant role |
| Depth | Active when using depth-based insertion; higher values place entry farther from the latest message |
Entry Toggle
In the Storybook list, each entry has a toggle switch. This switch only controls whether the entry is in an active state.
- On: The entry is on standby. Whether it's actually sent to the AI depends on its trigger mode (Constant entries are sent immediately; Keyword entries wait for a match).
- Off: The entry is fully disabled and will not be sent to the AI regardless of trigger conditions.
Trigger Modes
Constant: The entry is always injected — no condition required. Best for core character rules, fixed world-building that should be in effect throughout the entire conversation.
Keyword: The entry is injected only when a matching keyword appears in the conversation. Best for supporting characters, specific location descriptions, and event backgrounds that should be loaded on demand — tokens are only spent when the relevant content appears.
Insertion Position
Entries can be inserted at different positions in the context, affecting how the AI prioritizes the information:
- Before character definition / After character definition
- Depth-based insertion: Combined with the depth value, places the entry at a specific position relative to the latest message
- Message role: Injected as system, user, or assistant — affects how the AI interprets the content
Global Settings
The Storybook provides global configuration options that act as default matching behavior for all entries. Individual entries can override these in their own settings:
| Setting | Description |
|---|---|
| Scan Depth | How many recent messages to scan for keywords |
| Case Sensitive | Whether keyword matching is case-sensitive |
| Whole Word Match | Whether keywords must appear as whole words, not partial matches |
| Recursive Scan | Triggered entry content also participates in keyword matching, which can chain-trigger other entries |
Use Cases
- World-building: Place names, organizations, historical events — only injected when mentioned, avoiding unnecessary token spend
- Supporting characters: Automatically add a character's detailed description when their name appears in conversation
- Scene descriptions: Auto-inject environmental atmosphere when a specific location is referenced
- Constant rules: Always-on behavior constraints, formatting requirements, or character limits
Hiding Entry Details
Creators can hide the details of an entry. Once hidden, users cannot view or edit the entry's content — the original settings remain fully under the creator's control.
The AI is unaffected and will continue reading and using these entries normally. Hiding only prevents users from seeing the content; it has no impact on the Storybook's actual behavior.
This is especially useful when you need to protect core world-building, prevent users from modifying key rules, or safeguard plot hooks buried in the Storybook.
Tips
- Using keyword triggers instead of setting everything to Constant can significantly reduce unnecessary token consumption.
- Recursive scanning is powerful but can cause multiple entries to chain-trigger — enable with care.