Talk With Code is an innovative tool specifically engineered to simplify the process of analyzing GitHub repositories. By accepting links directly to the repository, it effortlessly deciphers the structure, providing a detailed list of all files present in the repository. The tool's unique feature is its ability to query not only code files but also common files like markdowns, thereby offering a comprehensive view of the repository. It further refines its analysis by asking users clarifying questions, such as the preferred programming language, ensuring customized and targeted results. Once the list of files is received, Talk With Code employs a relevance-based approach, iteratively querying the most relevant files until a relevance score of over 0.7 is achieved. This meticulous process ensures accurate and highly relevant results, making Talk With Code an invaluable asset for anyone seeking to gain in-depth insights into GitHub repositories.
Authorize me to access my GitHub repositories.
Show me an overview of my repo called "project
Can you list the files in the "master" branch of my "website" repository owned by "janedoe"?
I want to see the content of "main.py" file in the "dev" branch of "python_project" repository owned by "alice".
Get the code of "index.html" in the "gh-pages" branch of "my-portfolio" repo owned by "bob".
Description for AI
This plugin analyzes the GitHub repository provided by the user and then tries to answer users questions related to this repository. It accepts the link to the repository from the user. The first step is to analyze the structure of the repository. When querying the structure of the repository provide the relevant file extensions to answer the users question. Use not only the extensions for the code files but also the extensions of some common files, like markdown and others. If not sure about the extensions, ask the user some clarifying question, for example, which programming language he wants to use. The response is the list of all files that are present in the repository. After receiving the list of files the most relevant files should be queried iteratively. Before giving an answer to the user, the relevance of the answer should be calculated, if it less than 0.7, then additional files should be requested. The process should be repeated until the relevance score is higher than 0.7.