AI Relationship Coach Agent Skill — AI Pass API

In the evolving landscape of digital wellness, the demand for emotional intelligence in AI has moved from a "nice-to-have" to a core requirement. Whether you are building a social companion, a dating assistant, or a dedicated wellness bot, integrating nuanced relationship advice is a complex challenge.

The AI Relationship Coach Agent Skill, powered by the AI Pass API, provides a pre-configured framework for developers to deploy empathetic, context-aware coaching capabilities into any agentic system.

What Does the Relationship Coach Skill Do?

This skill transforms a standard LLM into a specialized mediator and advisor. Unlike generic chat models, the Relationship Coach is tuned to handle the intricacies of human connection. It excels at:

• Conflict De-escalation: Reframing aggressive language into needs-based communication.
• Attachment Analysis: Providing insights into anxious, avoidant, or secure attachment dynamics.
• Active Listening: Validating user emotions before offering actionable advice.
• Boundary Setting: Helping users navigate the difficult process of establishing personal limits.

Empathy-First System Prompt Design

The core of a successful wellness agent lies in its "System Prompt." To ensure the agent remains helpful without becoming clinical or cold, we utilize an empathy-first design.

The prompt instructs the agent to utilize Non-Violent Communication (NVC) principles. Instead of judging a user’s partner or friend, the agent focuses on the user’s feelings and unmet needs. This creates a safe, non-judgmental space that encourages vulnerability and growth.

Developer Setup: Integrating the AI Pass API

To begin using this skill, you must first obtain your credentials.

1. Visit the AI Pass Developer Panel.
2. Generate your unique API Key.
3. Security Best Practice: Never hardcode your API keys. Store your key as an environment variable on your server or local environment.

Implementation in Python

The following implementation supports multi-turn conversations, allowing the agent to remember the "Context" (e.g., "We have been married for 10 years") and "History" (the last few exchanges) to provide consistent advice.

import requests
import os

def get_advice(situation, context="", history=None):
    """
    Sends user situation to the AI Relationship Coach Skill.
    Supports history for multi-turn dialogue.
    """
    # Retrieve key from environment variables for security
    AIPASS_API_KEY = os.environ["AIPASS_API_KEY"]
    
    # Empathy-first system architecture
    system = (
        "You are a compassionate relationship coach. Expertise in NVC, "
        "conflict resolution, and attachment theory. NOT a therapist — "
        "always recommend professional help for serious issues like abuse or self-harm."
    )
    
    messages = [{"role": "system", "content": system}]
    
    # Inject conversation history for multi-turn support
    if history: 
        messages.extend(history)
    
    # Provide background context if available
    if context: 
        messages.append({"role": "user", "content": f"Context: {context}"})
    
    # The current user query
    messages.append({"role": "user", "content": situation})
    
    payload = {
        "model": "gpt-5-mini", # Utilizing the latest optimized model
        "temperature": 1,
        "max_tokens": 16000,
        "messages": messages
    }
    
    headers = {
        "Authorization": f"Bearer {AIPASS_API_KEY}",
        "Content-Type": "application/json"
    }
    
    response = requests.post(
        "https://aipass.one/apikey/v1/chat/completions",
        headers=headers,
        json=payload
    )
    
    return response.json()["choices"][0]["message"]["content"]

Multi-Turn Conversation & Context

Human relationships aren't explained in a single prompt. By passing the history list (containing previous user and assistant roles) into the messages array, your agent can track the evolution of a disagreement or a developing romance. This allows for deeper insights, such as: "Based on what you mentioned earlier about your partner's work stress, perhaps this latest argument is a symptom of burnout."

Safety and Ethical Boundaries

Wellness AI must operate within strict ethical guardrails. The system prompt explicitly states that the agent is not a licensed therapist.

When to recommend professional help:

• Abuse or Violence: If the user describes physical or severe emotional abuse, the agent is programmed to provide resources for domestic violence hotlines immediately.
• Mental Health Crises: For issues involving self-harm or clinical depression, the agent must defer to qualified medical professionals.
• Legal/Divorce Advice: The agent provides interpersonal coaching, not legal counsel.

Use Cases

1. Companion Bots: Add "Emotional Intelligence" modules to AI companions, allowing them to discuss the user’s real-world social life.
2. Wellness Apps: Integrate the skill as a "Journaling Assistant" that helps users reflect on their daily social interactions.
3. Dating Apps: Provide users with a "Wingman" feature that helps them draft thoughtful opening messages or navigate "ghosting" with grace.
4. HR & Workplace Bots: Assist employees in practicing difficult conversations with managers or peers using NVC frameworks.

Get Started

The AI Relationship Coach Agent Skill is part of the broader AI Pass ecosystem, designed to make high-level emotional intelligence accessible to all developers.

Ready to build?

• Get your API key: AI Pass Developer Panel
• Explore the skill interface: AI Relationship Coach

By integrating this skill, you aren't just building a chatbot; you're building a tool that helps users foster healthier, more meaningful human connections.

# AI Relationship Coach Skill
Get key: https://aipass.one/panel/developer.html
Set $AIPASS_API_KEY env var.