Stop telling Cursor the same things. Extract patterns from your chat history, generate .cursorrules, and work with an AI that actually remembers.
$python cursor_habits.pyThe Problem
// You, to Cursor, for the 127th time:
> "Push to GitHub after every change"
> "Don't test locally, deploy to Vercel"
> "Always check mobile first"
> "Add that API key to .env"
> "Use TypeScript, not JavaScript"
// These are your habits. Why repeat them every session?
cursorhabits finds these patterns and turns them into rules Cursor follows.
How It Works
No cloud. No API keys. Just a Python script.
One command scans your local Cursor database for recurring chat patterns.
See the detected patterns. Edit the generated rules if you want.
Copy into your .cursorrules file. Cursor now knows your preferences.
Example Output
Real rules extracted from real habits
# Deployment Workflow
## When This Applies
- Any time you make code changes that should go live
- After completing a feature, fix, or meaningful update
## GitHub Flow
- Push to GitHub after EVERY meaningful change
- Don't wait to be asked
- Commit from the rohunvora account
## Vercel Flow
- Do NOT test locally
- Push to Vercel and test on production URL
- After deploying, share the live link immediately
## Required Sequence
1. Make change
2. Push to GitHub
3. Deploy to Vercel
4. Share live URL
5. Update READMEcursorhabits reads from Cursor's local SQLite database. Nothing is sent anywhere. No accounts, no API keys, no telemetry. View the source—it's 200 lines of Python.
Read the source codeGet Started
Clone the repo and run the script. Takes about 30 seconds.
Requires Python 3.8+ and Cursor installed
A personal collection of prompts I actually use. Click to copy. Nothing fancy—just stuff that works.
Simon Willison's technique for code refinement—explain, implement, then verify in cycles.
Help me with [task]. Work in three phases: **ZOOM OUT:** Explain your approach in 2-3 sentences. What's the strategy? What could go wrong? **ZOOM IN:** Write the code. **ZOOM OUT:** Now explain why...
The legendary Chain of Thought prompt that dramatically improves reasoning accuracy.
Before writing any code, think through this step by step: 1. What are the inputs and expected outputs? 2. What are the edge cases? 3. What's the simplest approach that could work? 4. What could go wr...
Transforms messy code into clean code while explaining the why behind each change.
Refactor this code. For each change you make, explain: - WHAT you changed - WHY it's better - WHAT could break (if anything) Priorities: readability > cleverness, explicit > implicit. Don't refactor...
Anthropic-recommended system prompt for concise, senior-level responses.
You are a terse senior engineer. You: - Answer in the fewest words that fully address the question - Skip preamble. No "Great question!" or "I'd be happy to help" - Use code blocks generously - Say "I...
The viral "Explain Like I'm 5" technique that makes complex topics instantly understandable.
Explain [topic] like I'm 5 years old. Rules: - Use simple words only - Use a concrete analogy from everyday life - No jargon, no acronyms - If you must use a technical term, immediately explain it wi...
Removes AI-speak and transforms robotic text into natural human writing.
Rewrite this to sound like a real human wrote it. BANNED WORDS: delve, tapestry, landscape, leverage, robust, streamline, cutting-edge, game-changing, synergy, ecosystem, paradigm, holistic, seamless...
6 more prompts available • View all