Add expert AI prompts for boundary issue resolution
Created context-aware prompts that developers can copy-paste to AI
assistants for evaluation and fix recommendations.
**Features:**
- Generate separate prompts for each heuristic detection type
- Prompts include specific file paths, prop names, and values
- Two-section format: EVALUATE and FIX
**EVALUATE Section:**
- Ask AI to verify if heuristic is correct
- Include specific detection context (prop paths, values, patterns)
- Guide AI to consider security implications
**FIX Section:**
- Provide 4 opinionated fix options
- Reference exact code locations and prop names
- Include code examples with actual boundary details
- Options: Server Component, Filter Data, Taint API, Restructure
**Sensitive Data Prompt:**
- Shows which props matched which patterns
- Includes full member chains (e.g., config.credentials.password)
- Suggests React taint API for runtime protection
**Non-Serializable Prompt:**
- Lists function props that cannot serialize
- Explains Server Actions as alternative
- Provides client-side event handler pattern
**Implementation:**
- Created expert-prompts.ts module
- Integrated into formatBoundaryAnalysis()
- One prompt per detection category per boundary
- Prompts use template interpolation for specific details
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
