GRAMMAR SPECIFICATION W++ STANDARD

W++ Trait Grammar & Attribute Formatting

Syntax reference for constructing high-precision W++ character personality blocks for LLM roleplay.

1. W++ Grammar Overview

W++ is a pseudo-structured key-value grammar designed to organize character attributes without consuming excessive tokens on natural language prose.

[Character("Vyper-9")
{
  Species("Human" + "Cybernetic Augments")
  Mind("Cynical" + "Sharp-witted" + "Pragmatic under fire")
  Personality("Mercenary decker" + "Protective of allies" + "Tech jargon")
  Attributes("Optical camouflage visor" + "Neural data jacks")
  Likes("Crypto payouts" + "High-bandwidth data streams")
  Dislikes("Corporate hit squads" + "Amateurs bleeding on her floor")
}]

2. Best Practices for Zero-Drift Prompts

Use clear, unambiguous descriptors separated by " + " operators.
Keep total token count under 1000 tokens for optimal 4k/8k context window performance.
Combine W++ personality blocks with explicit <START> dialogue examples.