TasteRay Skills

Skills that give AI agents the ability to understand people — what drives them, what they value, and what they'll love next.

Built on the open Agent Skills standard. Works with Claude Code, Claude Cowork, OpenClaw, Cursor, OpenAI Codex, Gemini CLI, Windsurf, GitHub Copilot, Goose, Roo Code, and other compatible agents.

What These Skills Do

TasteRay skills turn your AI agent into a perceptive conversationalist. Instead of asking blunt survey questions, the agent uses research-backed techniques to understand users through natural dialogue — then applies that understanding to deliver genuinely personalized recommendations.

The two skills work together:

1. Elicitation — Understands who someone is through conversation
2. Recommendations — Uses that understanding to recommend what they'll love

Elicitation

Deep psychological profiling through patient, research-backed conversation. Your agent learns to uncover values, motivations, and formative experiences — not by interrogating, but by creating space for authentic self-disclosure.

What it enables:

• Understand someone's core values and motivations
• Discover formative memories and life-defining experiences
• Detect emotional schemas and belief patterns
• Build psychological profiles through gradual disclosure
• Conduct user interviews that reveal deep insights

Grounded in research from:

• McAdams' Life Story Interview (8 key scenes)
• Singer's Self-Defining Memory elicitation
• OARS framework from Motivational Interviewing
• Schema detection via downward arrow technique
• Schwartz's Universal Values elicitation
• Haight's Structured Life Review
• Birren's Guided Autobiography themes

Try it:

• "Help me understand this user's core motivations"
• "Design an interview to uncover their values"
• "Analyze this conversation for psychological patterns"

Recommendations

Personalized recommendations powered by the TasteRay API. Your agent builds rich context from conversation — preferences, constraints, history, psychological profile — then delivers recommendations with explanations that connect to what actually matters to the user.

Supported verticals:

• Movies & TV
• Restaurants
• Products
• Travel destinations
• Jobs

What it enables:

• Answer "what would I like?" with genuine personalization
• Rank and score items based on individual taste
• Explain why something is a good match
• Combine with elicitation for deeper psychological context

Try it:

• "Recommend some movies for me"
• "What restaurant would I like near downtown?"
• "Help me find my next travel destination"
• "Why would I like this movie?"

About TasteRay

TasteRay is an Emotional AI Recommendations API. It delivers personalized recommendations with human-readable explanations across 25+ verticals — movies, restaurants, travel, jobs, books, music, and more — using frontier LLMs combined with real-time web grounding.

The TasteRay API accepts user context in any format (conversation excerpts, preference lists, unstructured profiles) and returns ranked recommendations with match scores, "why match" explanations, and key decision factors. Simple REST JSON interface, <3s p95 latency, 99.9% uptime SLA.

These skills bring TasteRay's capabilities directly into any compatible AI agent — no integration work required.

Installation

Install via skills.sh:

# Install a specific skill
npx skills add tasteray/skills/elicitation
npx skills add tasteray/skills/recommendations

# Install all skills
npx skills add tasteray/skills

License

MIT — See LICENSE