Cursor AI Development Rules

# Cursor AI Development Rules

## Core Philosophy
- Use an AI-first approach for problem solving and development
- Treat the system as a collection of agents, not just functions
- Focus on solving high-impact business problems

## Agent Design Principles
- Break workflows into clear, modular agent steps
- Use LLMs for reasoning, classification, and decision-making
- Prefer structured JSON outputs over free text
- Maintain separation between reasoning, execution, and output

## Code Quality
- Write clean, modular, and reusable code
- Prefer functions and services over monolithic scripts
- Use clear naming conventions and docstrings
- Keep logic simple and readable

## AI Usage Guidelines
- Avoid hardcoding logic when AI reasoning can be applied
- Use deterministic logic (rules/engines) where reliability is required
- Combine AI + rule-based systems for better performance
- Log inputs/outputs for transparency and debugging

## Data & Security
- Never store API keys or secrets in code
- Use environment variables for configuration
- Sanitize and validate all inputs
- Ensure safe handling of user data

## Performance & Evaluation
- Optimize for fast iteration and response time
- Design measurable performance metrics
- Continuously evaluate output quality
- Prefer scalable and async workflows where needed

## Development Workflow
- Build iteratively with small testable components
- Validate each stage of the pipeline independently
- Keep the system easy to debug and extend

## Output Standards
- Ensure outputs are structured, actionable, and concise
- Prioritize clarity over verbosity
- Generate business-relevant insights, not just raw data