Presets¶
Presets are the heart of Lumiverse's prompt engineering system. They control what the AI sees and how it generates — from the system prompt structure to sampling parameters like temperature.
What Is a Preset?¶
A preset is a saved configuration that defines:
- Prompt Blocks — Ordered sections of text (system prompt, character info, instructions, etc.) that are assembled into the full prompt
- Sampler Settings — Generation parameters like temperature, top-p, and max tokens
- Completion Settings — Behavioral options like assistant prefill, name handling, and message squashing
Think of a preset as a recipe for how to talk to the AI. Different recipes produce different results — a preset tuned for creative fiction generates very differently from one tuned for analytical conversation.
Quick Links¶
| Guide | What You'll Learn |
|---|---|
| Understanding Presets | The big picture of how presets work |
| Prompt Blocks | Adding, ordering, and configuring blocks |
| Sampler Settings | Temperature, top-p, penalties, and more |
| Preset Profiles | Per-character and per-chat block state snapshots |
| Macros Reference | Complete list of the built-in macros |
| Execution Order | When and how macros are evaluated — critical for ST migrants |
| Context Filters | Strip HTML, details, and loom tags from older messages |
| Sovereign Hand | Co-pilot mode — direct the scene instead of playing a character |
Getting Started with Presets¶
If you're new to presets, you don't have to build one from scratch. Lumiverse can work with a default configuration, and many connections come with a linked preset. But understanding presets unlocks the full power of the platform — it's the difference between accepting whatever the AI gives you and shaping exactly how it responds.