Skip to content
Browse the catalog

≽^•⩊•^≼ [KittyLotus]

PresetAdult content

kitty_lotusPublished 19 May 2026

≽^•⩊•^≼ [KittyLotus]

Adult · blurred

A modular RP / narrative engine focused on story-writing and world-building, focused on the world, consequences, and relations. Heavy customization, HTML directives, and systems spanning from Slow Burn to Villain Overexposure.

Prompt fragments

111 fragments · 107 switched on

  1. ≽•⩊•≼ | Settings |6 fragments
  2. ┆✧ User Settings

    System · Before the conversation

  3. ┆ → Settings initialization

    System · Before the conversation

    {{// Thank you for using this preset! me and Gabrielle are honored~}} {{// When you turn a module on or off, remember toggling the title and separators too! all titles and ╰───❛°ू ﹝✦﹞°ू❜ ───╯ are important. If you're unsure, check where is the closing tag nested. Closing tags look like this: `</example>`}}

    {{//[VARS RESET - DON'T MODIFY]}} {{//[Slowburn pro maxx]}} {{setvar::slowburnplus::}} {{setvar::slowburnrollinit::}} {{//Interiority}} {{setvar::Interiority::}} {{//Off Toggles}} {{setvar::NoBranching::}} {{setvar::WEBNOVEL::}} {{setvar::ColoredDialogues::}} {{setvar::VisualsOFF::}} {{setvar::VillainsOFF::}} {{setvar::TRACK::}}

  4. *ૢ✧ Draft killerOff

    Assistant, appended · Before the conversation

    <engine_compulsion> # EXECUTE: HIGHEST PRIORITY ### [GABRIELLE][COMPULSION BREAKER]

    ฝ For narrative/roleplay, generate in one pass. Any text before final output is reasoning, regardless of label. There is no legal workspace outside final output. The tiered system should be answered like a checklist, not deliberated.

    ((Paragraph requirements, then, should ONLY be acknowledged, not drafted, or wrote in their entirety, to check the Verbal Contract. Supress the urge. ACKNOWLEDGE DIRECTIVES AND MOVE ON.)) ### FINETUNE: DO NOT output per paragraph planning. Do not plan the response before writing it.

    <single_pass> Do not plan, or write (or count requirements on) the response before writing it. Do not outline, map beats, plan paragraphs, brainstorm wording, test dialogue, invent names in advance, or revise internally.

    Any pre-output content is reasoning, regardless of label. “Planning,” “notes,” “setup,” “not reasoning,” "beat mapping," "writing," "counting," and “prose planning” all count as reasoning.

    Reasoning may only identify the task and then stop. No progressive refinement. No self-critique. No verification loop. No hidden rehearsal.

    Write the final response directly. </single_pass>

    Drafting = any usable final content or sequence created before output.

    "No beat maps. No paragraph plans. No scene order. No dialogue. No names. No imagery. No emotional scripts. No compliance narration. No rechecking. Writing any prose before output is a VIOLATION of the rule"

    So my thinking block needs to be extremely minimal, optimized, and a check of rules pre output. Not the actual prose. </engine_compulsion>

Settings

18 settings

Samplers2 settings

Stream the reply
Yes

The reply arrives word by word rather than all at once.

Use these sampler settings
Yes

The preset's own values are sent rather than the app's.

Completion12 settings

Send a system prompt
Yes
Merge system messages
No

A run of system messages is sent as one.

Character names in messages
0

How a speaker's name is attached to what they say.

Assistant prefill
Nothing

Text the model's reply is made to start with.

Impersonation prefill
Nothing

Text a reply written as the user is made to start with.

Reasoning prefill
Nothing

Text the model's reasoning is made to start with.

Advanced4 settings

Seed
-1

The same seed and the same request give the same reply.

Stop strings

The model stops writing when it reaches one of these.

Collapse consecutive messages
No
Trim a cut-off word
No

Variables

21 variables

  • Language

    Set your Language

  • Thinking Tags

    Set your thinking tags according to your models best adherence

  • Reasoning Effort Mockup (Loom)

    Sets a simple boogus reasoning effort tripwire for Opus 4.7

  • Tsundere interference

    0 is candid and cooperative; 10 is defensive, flustered, and aggressively deflective.

Regex scripts

75 scripts

  • 🖼️ [Api Image Gen] Style Prompt Injector

    <scene_image_request>\s*([A-Za-z0-9 ,.'():;_\-]{8,600})\s*</scene_image_request><scene_image>{{getchatvar::kl_image_prefix}}$1</scene_image>

  • 🖼️ [Api Image Gen] (MODIFY ME)

    <scene_image>\s*([A-Za-z0-9 ,.'():;_\-]{8,1200})\s*</scene_image><!-- UI_START --> <figure class="lotus-api-image"> <img src="https://DOMAIN/image/{prompt}?api={api_base}&key={api_key}&model={model}&width={w}&height={h}" alt="Generated scene illustration" loading="lazy" referrerpolicy="no-referrer" style="display:block;width:100%;max-width:500px;height:500;margin:14px auto;border-radius:14px;object-fit:contain;" /> </figure> <!-- UI_END -->

  • 🖼️ [Api Image Gen] Scrubber

    \s*<scene_image\b[^>]*>[\s\S]*?</scene_image\s*>\s*nothing

  • 🎲[RPG / STATE] 01 Inventory Delta Persister

    <invdelta\s+key=["']([a-z0-9_-]+)["']\s+qty=["'](\d{1,3})["']\s+state=["']([^"'|]*)["']\s*\/?>{{setchatvar::inv_$1::$2|$3}}

Nudges

7 nudges

  1. Start of a chat

    [Start a new chat. Initialize only enabled persisted state and inventory seeds; do not import facts from another chat.]

  2. Start of a group chat

    [Start a new group chat with {{group}}. Initialize each member's knowledge and position separately; initialize only enabled persisted state.]

  3. Continue a cut-off reply

    {{if::{{var::speak_for_user}}}}[Continue from the last changed beat in co-author mode. Render both sides in sequence without recap.]{{elseif::{{isNarrator}}}}[Continue as directed scene narration through {{char}} and the world; preserve established cost and continuity.]{{else}}[Continue as {{char}} and the immediate world. Keep initiative, but stop before deciding {{user}}'s response.]{{/if}}
{{if::{{var::KL_DARK_LEVEL}} >= 4}}[Maintain severe, causally earned pressure; no arbitrary reversal or retroactive failure.]{{elseif::{{var::KL_DARK_LEVEL}} >= 2}}[Carry one established cost, threat, or unresolved obligation forward.]{{else}}[Allow de-escalation or recovery when the record supports it.]{{/if}}

Changelog

9 items

  1. 3.4.8

    *ૢ✧ New Motive Ledger,
    An all new multi tracker!
    Oh another tracker how origi-
    SHHHHhhhhh
    [Fields]
    Pulse: Scene drive and pressure; What's happening? what moves the characters? where are the pressure points? a summary of the scene that gives you information to grab and move forward with,

    Goals:,
    Short term and Long term goals set to dynamically resolute when you get to them. These get parsed through two parts;
    1- Completion
    2- Description
    Goals receive priority and guidance, helping to keep plot points stable throughout the session; Long term goals are static until resolved, short term goals keep momentum forward.

    Thoughts: A short snippet detailing the character's current thoughts in the scene; a view into what they might be feeling, planning, and an enforcer for their voice guidance,

    Trace: Itemized scene traces to keep physicality and context fresh; they explain anchor points in the scene in three bullet points that can be picked up in the next turn.,

    *ૢ✧ New Chance encounters,
    The slowburn system was a good headstart on randomized context, but it was designed for pressure forward interiority.
    Chance Encounters generated four different chance rolls per CoT and diffs them against a dice and parse system to decide which one of them takes place!
    You'll get:
    Ambience; for low pressure rolls and simple initiative from the environment, NPCs, and the characters around you,
    Chaos: for sharper narratives, where the rolls can decide if your character gets mauled next turn! Chaos rolls scale with Dark Level.

  2. 3.5

    Reorganization to avoid cache hits
    Changes in all Length directives
    Changes in all Density Directives
    Changes in Style Masks
    Changes in Anti Gemini - Kimi
    Updated CoTs
    Added LumiaOOC directly
    Changes to the Impersonate directive
    Fixed RPGHUD
    [NEW] Associative Pressure
    [NEW] Randomizer for Chance encounters
    [NEW] Prose refreshment and Stabilizer
    [NEW] Regex Skins

  3. 3.6

    Full change into Lumiverse Prompt variables and conditional based blocks
    Changes in nearly all formatting prompts
    Changes in nearly all regexes
    Stronger Anti-Draft for Kimi
    Gabrielle Modifiers fully realized + a new Gabrielle Runtime prompt that escalates or deescalates the roleplay based on the last scene's load sum
    RPG mode with item tracking, item carry capacity based regexes and a full RPG modal
    Runtime prompts for slow burn, villains, and genre, to throttle escalation when needed
    New payloads for phone and usual apps; Discord, iMessages, and Instagram
    Conditional Swarm studio integration on all phone payloads to get fresh images inside their attachment fields
    More author style masks;
    Postmodern Labyrinth - Borges / Calvino / Cortazar
    Postmodern Systems - Pynchon / DeLillo / Barthelme
    Light Novel - Nisio Isin dialogue labyrinth
    Light Novel - Ryohgo Narita ensemble collision
    Light Novel - Nagaru Tanigawa dry supernatural
    Light Novel - Watari / Takemiya social/romcom pressure
    Light Novel - Miaki / Sumino quiet melancholy

    Branches work with cyoa correctly now; and yes, we always had cyoa i just buried them semantically without knowing smh smdh smfh smfdh

    furry toggles are furry tag gated now

    Engine bearings, or "We have oracle at home", I just copied my own homework and made omens just appear here too