AI Generation

This page is still work in progress.

The Story plugin uses LLMs (Large Language Models) to procedurally generate immersive content that adapts to the player’s world, context, and decisions. Rather than relying on static scripts or prewritten dialogue, Story leverages AI to simulate living, breathing worlds where every character has something to say, and every event has weight.

What Story Generates

  • Character Dialogue Conversations are not pre-scripted — they are generated live based on who is speaking, their context, and what they know or feel.

  • Memory Summaries After conversations, Story condenses exchanges into meaningful summaries stored in the character’s memory.

  • Lore Reinterpretation Characters recall or reference lore dynamically, influenced by their knowledgeCategories and personality traits.

  • Internal Monologue & Thought (planned) Story will support internal reflections from NPCs, especially in solitary moments or cinematic sequences.

  • Event Reactions When a major event occurs (e.g., war, betrayal, natural disaster), NPCs respond appropriately based on personality, allegiance, and relationship context.

Model Usage & Structure

Story sends structured prompts to LLMs containing:

  • Character Context: name, personality, memories, relationships

  • Location & World Data: current location, known events

  • Ongoing Scene: recent conversations or game events

  • Lore Hooks: relevant keywords tied to world lore

The model returns text which is then:

  • Parsed

  • Displayed via proximity chat or other dialogue systems

  • Stored in character memory (optional)

Last updated