Context-aware code assistant.
Make sure to add this to your project's package.json. Make sure to change default
"contributes": {
"configuration": {
"title": "Codextranaunts",
"properties": {
"codextranaunts.githubUsername": {
"type": "string",
"default": "",
"description": "Your GitHub username, this is purely used as an identifier for now, feel free to use anything"
},
"codextranaunts.repository": {
"type": "string",
"default": "",
"description": "Your GitHub repository, this is purely used as an identifier for now, feel free to use anything"
}
}
}
},
- Project description: https://docs.google.com/document/d/1sW52jTz1u8JYyERZfIxBgnt8UI3VW8VAGdCSBqK3ssw/edit
- Trello board: https://trello.com/b/v0vboA1o/pinecone-2023
- Architecture doc : https://docs.google.com/presentation/d/1FmQDaLUFrAJuS73uY0iG9flT2f5z1ZZSOOSxMshvFDQ/
Developers often receive bad answers when they employ tools like ChatGPT to aid in code debugging, refactoring, or feature addition. The current process requires a copious amount of copy/pasting to provide a little bit of context to the assistant. This approach does not allow the AI to fully comprehend the project's details, often leading to mismatched responses.
We propose the integration of a Large Language Model (LLM) to analyze and understand the code and readme.md files of a GitHub repository. This approach allows users to request assistance in debugging, refactoring, and expanding their project. The LLM retains all project information, giving it the ability to deliver the most appropriate and insightful responses. It has access to a lot of information, including additional files, tests, readme.md files, and more.
The following is a high-level outline of the proposed process:
- The LLM will scan and analyze the project's code and readme.md files in a GitHub repository.
- It will compile a context of the project using the obtained data and store it into Pinecone.
- When a user asks for assistance, the LLM will use this context to deliver a highly relevant response as well as a personalized prompt.
- The LLM will keep the context updated as the project evolves, ensuring the relevance of its assistance.
The exact product cost will depend on a detailed cost analysis, including factors such as infrastructure, model training, maintenance, and updates. However, a subscription-based model could be considered for monetization. Enterprises or individual developers could pay a monthly or annual fee for the advanced features and capabilities provided by this context-aware code assistant.
The potential real-world impact of this solution is substantial. By creating a context-aware AI assistant, developers can significantly enhance their productivity and improve code quality. It can save hours spent on debugging and refactoring, thus accelerating project completion timelines. Furthermore, it fosters a better understanding of the project among team members, leading to more robust and efficient software development processes.