Preset and Prompt Manager in Tavern Studio

A preset is where model behavior belongs. It can hold system prompt text, generation parameters, context limits, and prompt ordering rules. Keeping these settings in presets prevents every character card from becoming a copy of the same model instructions.

Tavern Studio treats presets and prompt assembly as first-class parts of the chat workflow. Character cards define who is speaking. World books define background knowledge. Presets define how the model should generate.

Who This Is For

  • SillyTavern users with existing presets.
  • Writers who switch between concise, descriptive, or dialogue-heavy styles.
  • Local LLM users tuning temperature, token limits, and context size.
  • API users who need different settings for different providers.

Core Content

Preset management is not about bypass tricks. It is about repeatable model behavior. A useful preset can include sampling settings, response length, context budget, system instructions, and ordering for prompt segments.

Tavern Studio's prompt assembly combines system prompt, character data, World Info, chat history, and generation settings before sending the request. For OpenAI-style APIs, this can be represented as structured message arrays instead of one plain prompt string.

How Tavern Studio Handles It

Tavern Studio supports preset import and local preset editing. Prompt order is part of the product logic: the assembler coordinates macro expansion, segment order, World Info injection, and context limiting.

This makes presets reusable. A character can be tested with one local model preset, then another cloud API preset, without rewriting the card.

Operation Steps

  1. Create or import a preset.
  2. Set the system prompt and generation parameters.
  3. Review context length and response token limits.
  4. Choose prompt order if the preset exposes it.
  5. Attach the preset to a chat.
  6. Send a test message.
  7. Adjust the preset instead of editing every character card.

FAQ

What is a preset?

A preset is a reusable model behavior profile: prompts, parameters, context limits, and ordering.

Is a preset the same as a character card?

No. A character card defines persona and scenario. A preset defines generation behavior.

Can I import SillyTavern presets?

Yes. Tavern Studio supports importing preset and JSON assets from compatible workflows.

Should prompts include unsafe bypass instructions?

No. Tavern Studio's preset workflow is for normal system prompts, formatting, style, and generation control.

Can different chats use different presets?

Yes. Different chats can use different presets and model connections.

Next Step