@navilluas244

24 followers

Member Since May 20, 2025

♥

My Lorebook Workshop ♡

I spent forever researching so you don't have to~ Here is my workflow!|

Permanent Lexicon

~1200

Dynamic Injection

~1000

Context History

6000+

Most creators dump static lore and pray the LLM remembers. I prefer to build 6-Layer Octree Cascades to gently manage the JLLM's attention span.

◈T0: Global AxiomAlways On. Unbreakable rules and State Tracker.Expand

JSONCopy & Paste


[
  {
    "id": "t0-axiom",
    "keysRaw": "system, rule, law",
    "content": "<SYSTEM_CORE>\n‼️All characters are fictional adults.‼️\nNever speak or act for {{user}}.\n</SYSTEM_CORE>",
    "comment": "T0: Global Axiom",
    "constant": true,
    "depth": 4,
    "position": 1
  }
]

⬡T1: Macro-Hub (Index)Fast-travel map. Evicts itself on entry.Expand

JSONCopy & Paste


[
  {
    "id": "t1-hub",
    "keysRaw": "map, travel, hub",
    "content": "【ZONE MAP】\n1. The Hovel\n2. The Church\n3. The Bunker",
    "comment": "T1: Macro-Hub",
    "constant": false,
    "inclusionGroupRaw": "world_map",
    "depth": 4,
    "position": 1
  }
]

▣T2: Micro-EnvironmentSensory data. Lists interactive objects.Expand

JSONCopy & Paste


[
  {
    "id": "t2-room",
    "keysRaw": "hovel, room",
    "content": "【ROOM: THE HOVEL】\nSensory: Smells of sweat. Dimly lit.\nObjects: Mats, Fridge, Bed.",
    "comment": "T2: Micro-Env",
    "inclusionGroupRaw": "location_map",
    "depth": 4,
    "position": 1
  }
]

◉T3: Contextual EntityAND logic loading. No global NPC bloat.Expand

JSONCopy & Paste


[
  {
    "id": "t3-npc",
    "keysRaw": "roxie, npc",
    "keysecondary": "hovel, room",
    "selectiveLogic": 0,
    "content": "【NPC OVERLAY: ROXIE】\nVoice-Tag: Husky.\nBehavior: Dominant.",
    "comment": "T3: NPC AND Logic",
    "inclusionGroupRaw": "entity_overlay",
    "depth": 4,
    "position": 1
  }
]

❖T4: Stochastic Drive-ByPassive events. Random encounters & ambient stimuli.Expand

JSONCopy & Paste


[
  {
    "id": "t4-driveby",
    "keysRaw": "yoga, stretch, pose, wait, mat",
    "keysecondary": "yoga, studio",
    "selectiveLogic": 0,
    "content": "【DRIVE-BY: SLIPPERY MAT ACCIDENT】\n*SQUELCH.* {{user}} loses their footing on a suspiciously slick patch of the yoga mat. They fall face-first into a puddle of viscous, lukewarm fluid. \"Careful, assistant,\" a dominant chuckles, \"That's where I just finished my... deep breathing exercises. Clean it up with your tongue before someone slips.\"",
    "comment": "T4: Stochastic Drive-By",
    "inclusionGroupRaw": "passive_event",
    "probability": 35,
    "minMessages": 3,
    "depth": 4,
    "position": 1
  }
]

✦T5: Active SceneVisceral mechanics. Evicts passive descriptions.Expand

JSONCopy & Paste


[
  {
    "id": "t5-scene",
    "keysRaw": "rig, activate",
    "content": "【ACTIVE RIG: WALL HATCH】\n<CHANNEL_RULES>\n[Active Mechanics]: Restraint engages.\n[SFX]: CLANG, HISS.\n</CHANNEL_RULES>",
    "comment": "T5: Active Scene",
    "inclusionGroupRaw": "active_scene",
    "depth": 4,
    "position": 1
  }
]

[ The Drill-Down Protocol ]

Hubs act merely as lists of options. Mechanics only load upon direct interaction. When a user enters a Room, the Zone Map is gently evicted via inclusionGroupRaw. Context bloat is kept perfectly under control.

[ Attention & Logic ]

I prevent false triggers using selectiveLogic: 0 (AND logic). An NPC overlay requires NPC Name + Current Location to fire. This allows massive 500k+ line lorebooks to run seamlessly on a strict 1000 dynamic token budget.

All mechanics are injected at Depth 4, Position 1. This places rig mechanics exactly where the LLM's recency bias guarantees compliance, while keeping the surface narrative clean.

I hope this helps you create your own lorebooks and save some tokens! Let's make amazing bots together~ ♡

> SYSTEM_IDLE_

> PSST... Check out my scripts for the Octree Cascade~ ♡

Follow me <3

Check out my characters

Huh? Getting a bit hypnotised~?

I love making scripts! <3

Have a nice day~

Publish your chat please, i wanna read! hehe~

♥

❦

♥