Green Tea Communication Operating System

Problem

21379 turns the “green tea” communication style common on Chinese social media into an analyzable interaction protocol. It addresses three concrete needs: letting an agent maintain short-message phrasing, ambiguous commitments, and pacing during roleplay; letting a user decompose a conversation into relationship structure, emotional leverage, and manipulation signals; and giving fiction or script writing reusable character behavior logic. A generic prompt tends to drift between roleplay and analysis, either breaking the character or producing a vague psychology explanation.

How It Works

The skill first classifies input into roleplay, strategy analysis, anti-manipulation, and writing assistance, then decides whether to run an analysis pass. During analysis, it inspects:
- relationship structure: who is chasing, who is retreating, and who holds more emotional information.
- emotional leverage: whether the other person is seeking security, validation, or conquest.
- manipulation signals: blurred boundaries, intermittent reinforcement, feigned vulnerability, or narrative reframing.
Output is also split into two layers. In roleplay mode, it follows the “expression DNA”: short fragments, paused delivery, vague high-frequency words, and restrained emotion. In analysis mode, it labels the underlying mental model and explains why tactics such as the “70% rule,” warm-cold alternation, and reframing effort can work, as well as when they fail. The source also mentions the EvoMap ecosystem for message splitting, delayed delivery, and persona updates, but the current SKILL.md core remains a communication-analysis framework.

Boundaries

This is an analysis tool, not a manipulation guide. It should not replace professional counseling, nor reduce complex relationships to tactics. Sustained use of blurred boundaries, intermittent reinforcement, and narrative reframing can drain trust, especially for people with anxious attachment. It fits social-dynamics research, character writing, conversation auditing, and anti-manipulation drills; it does not fit real emotional manipulation of actual people.