This repository provides three straightforward prompts designed to significantly enhance the performance of your LLM coding agents. These client-side prompting techniques, explained comprehensively in the linked article, ensure better accuracy and quality in your coding tasks.
- Plan: Guides your agent to thoroughly analyze the task and create a detailed, step-by-step solution plan.
- Code: Directs your agent to implement the previously developed plan effectively and accurately.
- Review: Allows your agent to review implemented code rigorously, identify potential issues, and make necessary improvements.
- Immediate Improvement: Boosts the effectiveness of any LLM agent instantly.
- Cost-Effective: Ideal for IDEs or services with flat-rate monthly subscriptions.
- Easy to Integrate: Simply copy the prompts into your existing workflow.
Use each prompt in a separate LLM conversation for optimal results:
- First, generate a detailed Plan.
- Next, execute this plan using the Code prompt.
- Finally, ensure quality with the Review prompt.
For the best results, consider adding an AI-README.md file to your repository to give agents essential context about your project structure and standards.
For a detailed explanation and further insights into the methodology, see the full article here.
Happy coding!