Personal Insurance Coverage Coach

AI-powered insurance assistant that transforms insurance shopping through personalized coaching, local agent connections, and professional carrier negotiation tools using conversational AI.

What This Skill Does

This skill converts a CLAUDE.md project context file from an insurance assistant application into actionable development guidance. The application provides:

Tech Stack Context

Step-by-Step Instructions

1. Understand Critical useChat Rules

**NEVER do these:**

**ALWAYS do these:**

2. Implement Information Collection Flow

When gathering insurance information:

1. **Check existing data FIRST** - never ask for already-known information

2. **Provide 3-4 specific numbered options** - never end with vague questions

3. **Follow insurance-specific order**:

- Auto: Drivers → Vehicles → ZIP → Ages → Vehicle Details → Coverage

- Home: Property type → ZIP → Home value → Year built → Coverage amount

- Life: Coverage amount → Health status → Smoking → Term vs whole life

4. **Require 5 mandatory fields for auto quotes**: drivers, vehicles, ZIP, ages, vehicle details

5. **Track completion percentage**: 0-40% (basic), 40-70% (core), 70-90% (most complete), 90-100% (ready)

3. Implement Quote Path Selection

After initial options, ALWAYS offer:

```

"Would you like:

A) Quick quote (5 minutes) - Basic information for rough estimate

B) Detailed quote (10-15 minutes) - Comprehensive info for accurate pricing"

```

**Quick path**: Minimum fields only, provide estimate ranges

**Detailed path**: Comprehensive gathering, provide specific estimates with carrier comparisons

4. Build Carrier Conversation Toolkit

Generate two sections:

**REQUIRED COVERAGES** (non-negotiable):

**FLEXIBLE COVERAGES** (negotiation opportunities):

Include negotiation scripts for each coverage type.

5. Handle Real-Time Profile Persistence

1. Extract data from ALL messages using regex patterns (age, vehicles, marital status, ZIP, home value)

2. Merge with existing localStorage profile

3. Save immediately after extraction

4. Auto-fill in system prompt for next response

5. Update UI via custom event listeners

6. Follow Project Structure

```

/app/api/chat - OpenAI endpoint

/components/chat-interface.tsx - Main UI

/lib/carrier-conversation-helper.ts - Toolkit generation

/lib/customer-profile.ts - Profile management

/lib/information-tracker.ts - Data extraction

/lib/rag/core/vectorize-io-client.ts - RAG integration

```

7. Apply Coding Conventions

8. Configure Environment

Required variables:

```

OPENAI_API_KEY - OpenAI GPT-4 access

VECTORIZE_IO_API_KEY - RAG system

VECTORIZE_IO_ORG_ID - Organization ID

VECTORIZE_IO_PIPELINE_ID - Pipeline ID

ENABLE_RAG=true - Enable RAG features

```

9. Test Implementation

Run these commands before committing:

```bash

npm run lint # ESLint checks

npm run build # Type checking

node test-openai.js # API connection test

```

10. Handle Insurance-Specific Logic

**Premium Ranges**:

**Always show three tiers**: State minimum, Standard coverage, Full coverage

Examples

Correct useChat Implementation

```typescript

const { messages, input, handleInputChange, handleSubmit, isLoading } = useChat({

api: '/api/chat',

body: { customerProfile },

onError: (error) => console.error('[Chat Error]:', error),

})

```

Proper Prompt Presentation

```

"I can help you with:

1) Get an auto insurance quote

2) Compare home insurance options

3) Explore life insurance coverage

4) Review your current policies

Which would you like to explore?"

```

Constraints

Notes