How TavernAPI Roleplay Presets Work
To use a TavernAPI preset, download the preset JSON from our console, import it into SillyTavern, and select any standard model ID (such as gemini-4-argon or claude-opus-5.5). You control prompt order and customizable toggles locally in SillyTavern, while our gateway replaces locked [[tav:pack/block]] placeholders with the core writing instructions on the server.
Why we keep prompt ordering in SillyTavern and lock core blocks on the server
A complex SillyTavern preset has two halves: dozens of optional toggles, regex bindings, and ordering slots you want to control yourself, and core storytelling instructions that took months to tune against AI clichés. Moving the whole preset onto the server would strip away your local toggles; distributing the core instructions in plain text means they get copied and leaked overnight.
Our solution splits the job cleanly: SillyTavern handles prompt ordering and your custom toggles, while our server holds the secret core text. In the preset file you download from our console, locked entries look like normal Prompt Manager rows—you can see their names, toggle them on or off, and drag them into any order—but their body contains a single placeholder line like [[tav:pack/block]]. When SillyTavern sends the assembled request to https://tavernapi.com/v1, our gateway replaces each placeholder with the full instructions, expands {{char}} and {{user}}, and sets prompt-cache breakpoints automatically. You don't need Tavern Helper or any plugins, and the preset only works through TavernAPI.
What you can customize locally vs. what happens on the server
When you import a TavernAPI preset into SillyTavern's AI Response Configuration panel, here is how every part of your setup behaves:
- Open customizable entries
- Toggles for response length (word count), output language, narrative perspective, and anti-impersonation (preventing the AI from speaking or acting for {{user}}) are plain text. Edit them, toggle them on or off, or add your own custom prompt rows anytime.
- Locked core blocks ([[tav:pack/block]])
- Core storytelling, character voice, and anti-slop entries display their names in your Prompt Manager list so you can toggle or reorder them, while their [[tav:pack/block]] placeholders are expanded on our server right before the model is called.
- Character card, Persona & Lorebook
- Because SillyTavern still builds the prompt locally, your character card description, personality, scenario, <user> persona, World Info / Lorebook entries (including @depth injections), Author's Notes, and regex scripts all work exactly as normal.
- Raw model connection without our preset
- Using our preset is completely optional. If you prefer your own local SillyTavern preset (or connect from another frontend like RisuAI or JanitorAI), just select the same model ID (like gemini-4-argon or claude-opus-5.5)—requests without [[tav:...]] placeholders pass straight through with automatic prompt caching.
- Standard Preset unlock rules
- Unlocked permanently on your account as soon as you complete your first top-up of $5 or more.
- Pro Preset unlock rules
- Available three ways: purchase it standalone for $4.99, use it for free while an active $19 or $39 30-day credit pack is in effect, or unlock it permanently once your cumulative spend reaches $25.
How to download and import a TavernAPI preset in 4 steps
Download your preset JSON from the Presets tab in the TavernAPI console and import it under SillyTavern's Chat Completion presets—no plugins or extra files required.
All signed-in users can browse and download the sanitized preset files from the console (they contain no secret text—our gateway checks your account's preset access when you send a message):
- Sign in to the TavernAPI console, open the Presets page in the left sidebar, and click Download on the preset pack you want.
- In SillyTavern, click the AI Response Configuration icon (the sliders on the far left) and click Import preset next to Chat Completion Presets.
- Select the downloaded TavernAPI-<name>-<version>.json file. Turn on the plain-text toggles (such as word count or language) that fit your story.
- In API Connections (the plug icon), connect to https://tavernapi.com/v1 and pick your model ID (such as gemini-4-argon or claude-opus-5.5) from the dropdown.