cursor-prompts 1.0.0

A local-first tool that captures and analyzes Cursor prompts for quality and effectiveness

Cursor Prompts

A local-first tool that captures and analyzes Cursor prompts for quality and effectiveness.

Installation

pip install cursor-prompts

Setup

1. Configure hooks:

   cursor-prompts setup
   

   This creates hooks in ~/.cursor/ and initializes storage in ~/.prompt-analyzer/.

2. Use Cursor normally - prompts are captured automatically.

3. Review your prompts:

   cursor-prompts stats              # View statistics
   cursor-prompts examples --type rejected  # See problematic prompts
   cursor-prompts recommend          # Generate Cursor rules and commands recommendations
   

Quick Commands

• cursor-prompts stats [--since 7d] - Show summary statistics
• cursor-prompts examples [--type rejected|repeated|all] [--limit N] - Show examples
• cursor-prompts recommend [--since 30d] [--project PATH] [--no-open] - Generate Cursor rules and commands recommendations from recent prompts
• cursor-prompts storage - Show storage info
• cursor-prompts storage clear [--older-than 30d] - Clear old prompts

Analysis Criteria

Prompts are scored 0-100 with deductions for:

• Rejected Suggestions (-30): AI response was rejected
• Repeated Prompts (-20): Same/similar prompt sent multiple times (>80% similarity within session or 5-minute window)
• Vague Requests (-15): Very short prompts or single-word questions lacking context

The tool provides contextual suggestions for improvement based on detected issues.

Recommendations

The recommend command analyzes your prompts to generate Cursor rules and commands recommendations:

• Global patterns: Identifies common patterns across all your projects
• Project-specific: Generates recommendations tailored to specific projects
• Existing detection: Scans for existing .cursorrules files and commands to avoid duplicates
• Interactive HTML: Opens a browser page with all recommendations, organized by project

Example usage:

cursor-prompts recommend                    # Analyze last 30 days
cursor-prompts recommend --since 7d        # Analyze last 7 days
cursor-prompts recommend --project /path   # Analyze specific project
cursor-prompts recommend --no-open         # Save HTML without opening browser

Storage

All data is stored locally:

• Database: ~/.prompt-analyzer/data/prompts.db
• Config: ~/.prompt-analyzer/config.json
• Hooks: ~/.cursor/hooks.json and ~/.cursor/hooks/cursor-prompts.js

Requirements

• Python 3.8+
• Cursor IDE
• Node.js (for hook execution)

Development

pip install -e ".[dev]"
pytest

License

MIT License