Search code, repositories, users, issues, pull requests...

AI models like ChatGPT/Claude/Cursor struggle to create accurate fullstack webapp from scratch. They are not bad at coding but get confused due to following reasons:-

We try to solve these problems by dividing the code generation process into distinct steps/phases where we force the AI to document and save all of its assumptions and subsequent code details.

To increase its accuracy we pass custome prompts and also pass the output of previous steps as context to generate the next response.

You just give it a simple prompt like "Create an expense management tool" and it will set up the whole project with relevant code and requirement docs. We explain the chart below on how it takes user prompt and creates the app:

Each phase uses specialized prompts that focus on specific aspects of the application to ensure the best output is generated.\n We also have different versions of prompts for you to try out and see which leads to highest accuracy code.

Create a .env file in the root directory with your API keys: env OPENAI_API_KEY=your_openai_key ANTHROPIC_API_KEY=your_anthropic_key

Leave a Comment