[New Feature]: prompt improvement for various best practices

[yunks128]

Checked for duplicates

Have you checked for duplicate issue tickets?

Yes

Alternatives considered

Have you considered alternative solutions to your feature request?

Yes - and alternatives don't suffice

Related problems

There is a noticeable performance gap between local language models from Ollama (llama3.1:8b, llama3.1:70b) and the cloud-based larger models from OpenAI and Azure.

Describe the feature request

I propose the implementation of more specific, tailored prompts for each best practice in order to enhance performance. Additionally, reference information that feeds into the model context should be customized for each best practice, ensuring that the model can provide more accurate and context-aware responses.

[yunks128]

Steps for improvement

• Break down problems into modular components (shorter inputs and outputs)
• Sequential organization for outputs (multiple iterations for final outputs)
• Extract metadata in target repo (e.g., for GOVERNANCE.md, we don't need entire repo content for input context; we will need only metadata as input)

[riverma]

@yunks128 - FYI there's a ticket where we can use a more seamless architecture to support this: #25

I'm thinking we have a Python class that each best practice implements, including implementing functions like 'apply', 'deploy', etc. AI prompts can go in each respective best practice's python class as well - possibly as a constant or array of prompts to choose from.

Also - for this ticket, let's be specific about which best practices you want to improve prompts for. Otherwise the scope for this ticket is too broad to ever close. Which best practices do you want to improve the prompts for?

[yunks128]

@riverma This issue no longer seems relevant. As you mentioned, issue #25 is addressing the re-architecture. In the meantime, I'm focusing on the generate-docs function in a separate repository: https://github.com/yunks128/docusaurus-generator