These core strategies apply to all AI coding assistants, ensuring you get accurate, high-quality, and contextual suggestions.
- Be Specific and Constrained: Replace vague requests with clear, specific, and detailed instructions (e.g., "Refactor this function to use async/await instead of promises and include proper input validation and TypeScript generics.").
- Define Output Format: Specify the expected output (e.g., "Generate a unit test using Jest for this function," or "Provide a class diagram in Mermaid format.").
- Iterate and Refine: Start with a simple prompt and incrementally refine your request based on the AI's previous output. Don't expect perfection in one go.
- Explicitly Provide Constraints: Always include contextual requirements such as the framework (e.g., React hooks, Spring Boot), specific libraries (e.g., Zod, Pandas), and existing patterns.
- Reference Related Code: Mention or include related files, functions, or class names that the AI needs to understand to maintain architectural consistency.
Example: Explain your goal: "I'm creating a new API endpoint. It must follow the same error handling and logging pattern as in the
UserService.tsfile." - Explain Intent/Business Logic: Tell the AI why the code exists or what the feature is meant to achieve, not just what to write.
- Review Before Committing: Never blindly accept generated code. Treat AI output as a draft that requires human review for logic, edge cases, and security.
- Test Generated Code: Prioritize testing (manual and automated) on AI-generated components, especially security-critical or performance-sensitive parts.
- Maintain Fundamentals: Use AI to accelerate your work, not replace your understanding. You must still be the domain expert and final arbiter of code quality.
- system-prompts-and-models-of-ai-tools
- awesome-cursorrules
- Rules template
- rulebook-ai
- LEAKED SYSTEM PROMPTS
- devin.cursorrules
- agent-rules
- awesome-cursor-rules-mdc
- Awesome Vibe Coding
- awesome-ai-coding-tools
- awesome-claude-code
- awesome-ai-system-prompts
- awesome-claude-agents
- awesome-claude-code-subagents
- context-engineering-intro
- claude-code-templates
- claude-code-security-review
- claude-code-guide
- claude-code-subagents-collection
- Mastering-GitHub-Copilot-for-Paired-Programming
- claude-code-settings
- how-to-build-a-coding-agent
- claude-code-infrastructure-showcase
- Spec-driven development