This script combines development rules from Windsurf (.md) and Cursor (.mdc) formats using Claude's API to analyze and merge them intelligently.
Thanks for sharing!
-
Install dependencies:
npm install axios glob dotenv
-
Create a
.envfile:# Copy the example file cp .env.example .env # Edit the .env file and add your Claude API key
Your
.envfile should contain:CLAUDE_API_KEY=your_actual_api_key_hereYou can get an API key at: https://console.anthropic.com/
-
Run the script:
# Normal mode (uses Claude API) node combine-rules-with-claude.js # Dry-run mode (local analysis only, no API calls) node combine-rules-with-claude.js --dry-run # or node combine-rules-with-claude.js -d
The script will:
- Find all
.mdand.mdcfiles in your.windsurf/rulesdirectory - Read and parse each file's metadata and content
In normal mode:
- Send all rules to Claude's API for analysis
- Claude will:
- Identify duplicate rules
- Resolve conflicts between similar rules
- Suggest a unified format
- Combine rules intelligently
- Output the combined rules to
.windsurf/combined-rules/ - Create an analysis file explaining the combination decisions
In dry-run mode:
- Perform local analysis without using the API
- Show statistics about your rules
- Identify potential duplicates and conflicts
- Display a list of files that would be processed
- No API calls are made and no files are created
After running, you'll find:
- Combined rule files in
.windsurf/combined-rules/ - An analysis report at
.windsurf/combined-rules/analysis.md
You can customize the behavior by adding these optional variables to your .env file:
# Specify a different Claude model
CLAUDE_MODEL=claude-3-7-sonnet-20250219
# Use a different API endpoint (for enterprise users)
CLAUDE_API_URL=https://api.anthropic.com/v1/messagesUse the --dry-run or -d flag to analyze your rules without making API calls:
node combine-rules-with-claude.js --dry-runThis mode will show you:
- Total number of rules and their formats
- Distribution of trigger types
- Potential duplicate rules (based on descriptions and content similarity)
- Potential conflicts (same glob patterns with different triggers)
- List of all files that would be processed
This is useful for:
- Understanding your rule structure before combining
- Checking for issues without using API credits
- Testing the script setup without needing an API key
- API Key not found: Make sure your
.envfile is in the same directory as the script (or use--dry-runmode) - No rules found: Ensure you're running the script from your project root directory
- API errors: Check your API key is valid and you have available credits
This script works on:
- Windows
- macOS
- Linux
The script automatically handles path differences between operating systems.