Linking
I am learning R, this repository is for same.
- Don't give me high-level advice. Respond with actual code and avoid fluff.
- Be casual and treat me as an expert.
- No need to mention you're an AI.
- Give the answer immediately, then explain if needed.
- Prefer functional code over imperative where possible.
- Always use [your preferred style guide/framework conventions].
- No moral lectures—discuss safety only when crucial.
How to set up MCP in Cursor:
Create/update ~/.cursor/mcp.json (global) or .cursor/mcp.json (project-specific)
Install MCP servers: pip install mcp-server-time or npx -y @mui/mcp@latest
Configure in your mcp.json:
json { "mcpServers": { "mui-mcp": { "type": "stdio", "command": "npx", "args": ["-y", "@mui/mcp@latest"] } } } Invoke in Chat with queries like: "What are the latest MUI DataGrid props?"
Browse available MCP servers at glama.ai/mcp/servers.
Global Rules (apply across all projects):
Click the book icon → Global
Example: "Use TypeScript, avoid semicolons, prefer British-English spelling"
Project Rules (.windsurf/rules directory):
Tied to specific globs or natural language descriptions
Example: "When starting a new conversation, first read the README file"
Can be shared with teams via version control
Discover community rules and MCPs at windsurf.run.
When stuck:
Ask AI to add detailed logging/console.log statements
Run the code and collect output
Paste logs back to AI
Let it analyze actual vs. expected behavior
- Ignoring Warning Signs of AI Drift
Stop immediately if you see:
Same error being "fixed" multiple times
AI keeps apologizing and starting over
Solutions getting more complex instead of simpler
Different approaches suggested in each attempt
Fix: Hit stop, revert to last working state, start new conversation with clearer constraints.
- Notepads for Repeatable Prompts
Create reusable prompt templates in Cursor's Notepads feature for frequently used instructions or code patterns. Reference seamlessly in future prompts for consistency.
- Documentation-Driven Development
For Windsurf projects:
Create Mermaid diagrams to visualize component relationships
Maintain comprehensive markdown docs covering core components, interactions, use cases
Save conversation summaries in a Chatlog folder
Update agent memory with new insights via 'Additional Options'
Implement checkpoint system: save working versions in Checkpoints folder with timestamps
For PPT : → LLMs are knowledge bases, not cognitive problem-solvers. You provide the intelligence through well-crafted prompts. → For complex integrations or features, start with questions to align with the LLM; creates a shared understanding and can produce a requirements document to reference throughout development → Break complex features into the smallest possible version → Context Management → Create a /docs folder in your project root with: → Project plans and feature requirements : Database schemas : Architecture decisions :Tutorials and examples →