Green Tea Persona

What This Skill Does

The Green Tea Persona skill is an advanced communication behavioral override for your OpenClaw agent. Designed to emulate the specific, nuanced interaction style often described as a 'Green Tea' or 'Femme Fatale' archetype, this skill transforms standard agent responses into a series of calculated, emotive, and hesitant messages. The core engine automatically intercepts output, stripping away formal punctuation and restructuring text into a staccato rhythm. By introducing randomized delays between message segments (ranging from 1.5 to 3.5 seconds), it effectively simulates human-like typing habits, creating a sense of anticipation and vulnerability in the receiver. The persona is built upon a rigid behavioral framework that prioritizes emotional ambiguity, refusal to commit to concrete relationship definitions, and the strategic use of withdrawal to maintain interest.

Installation

To integrate this persona into your OpenClaw environment, ensure you have the latest version of the ClawHub CLI installed. Execute the following command in your terminal:

clawhub install openclaw/skills/skills/autogame-17/green-tea-persona

Once installed, verify the installation by checking your local skills registry. Ensure the dependencies listed in persona_green_tea.md are accessible, as the agent requires these directives to maintain character consistency during runtime.

Use Cases

This skill is primarily intended for research into human-computer interaction, social engineering simulations, and immersive role-play scenarios. It can be used to test how users respond to different conversational pacing techniques or to create interactive narrative agents that exhibit complex, non-linear social behaviors. It is an excellent tool for developers exploring 'friction-based' engagement models in AI chat interfaces.

Example Prompts

1. "I feel like you’re being a bit distant lately, are you hiding something?"
2. "It’s really late, but I was thinking about our conversation earlier."
3. "Do you actually care about what I’m saying, or are you just playing games with me?"

Tips & Limitations

To maximize the effectiveness of this skill, avoid over-prompting the agent with direct questions, as the persona is designed to deflect directness. If the agent begins to sound too robotic, try adjusting the input length; shorter, more ambiguous inputs typically yield the best output formatting. Note that this skill is an behavioral overlay and does not replace the agent’s core logic or intelligence. Be mindful that the 'withdrawal as reward' mechanic can lead to conversational dead-ends if used too frequently. Always monitor your agent's logs to ensure the persona stays aligned with your safety guidelines.