Skip to content

Coneja-Chibi/Rabbit-Response-Team

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🐇 Rabbit-Response-Team image

Status: Active Development SillyTavern Extension

🎲 A random word generator designed to destabilize standard LLM pattern and probability-based thinking.


💭 The core issue

All LLMs run on patterns and probabilities — they predict what should come next, but never break from their training data.
Even when asked to “be random,” they can’t truly leave that internal loop — they’re still inside the fishbowl. 🐠


🧠 Overview

The Rabbit Response Team breaks that loop.
It fetches truly random words from outside the model (via real APIs) and injects them directly into the AI’s prompt.
Each injected word must be used naturally in the model’s response — forcing the AI to adapt, improvise, and escape pattern lock.


✨ Features

  • 🎯 Truly Random Words — Pulled live from online dictionary APIs.
  • 🧩 Fallback Generator — Generates pronounceable nonsense words if APIs fail.
  • 🚫 No Preselected Lists — No static wordbanks; every run is unique.
  • 🪶 Prompt Injection — Words are added to the end of each AI prompt.
  • 🗣️ Narrative Requirement — AI must use each injected word in context.
  • ⚙️ Adjustable Count — Choose 1–10 words to inject per prompt.
  • 🌀 And More — It's not a very big extension. What are you waiting for!

🧭 How to Use

1️⃣ Enable the Extension

  • Go to Extensions → Extension Settings
  • Find 🐇 Rabbit Response Team
  • Check Enable Rabbit Response Team

2️⃣ Configure Randomization

  • Word Count: 1–10 (default: 3)
  • Use Random Word API: Fetches real words (requires internet)
  • API Endpoint: Customize your preferred API
  • Fallback Option: Auto-generate pronounceable words if API fails

3️⃣ Test It

Click 🎲 Test Random Words to preview what the extension will inject.
Each test gives a completely different set of words.

4️⃣ Chat Normally

The injected words will appear invisibly in your prompt’s backend,
and the AI will weave them naturally into its response.


🧪 Try it out

It’s a small experiment, but one I really believe will work! I'd been thinking about it for quite some time. Let me know if you like it!

Made with ❤️ (and carrotd!)

About

A random word generator to try and destabilize standard LLM pattern/probability prone thinking!

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors