-
Home screen, information about the application and entry fields for either single-file submission, or authorization through GitHub
-
While processing a request, guards are in place to prevent erroneous submissions while one is currently in progress.
-
After a submission is processed, generated markdown is displayed, allowing the user to copy/paste or download their documentation
-
After authentication, API routes recursively search through a user's public and private repositories, showing available files and folders
Inspiration
The inspiration for docuMint arose from the tedious and time-consuming task of creating and maintaining code documentation. We recognized the need for an automated solution that could simplify this process, allowing developers to focus more on coding and less on documenting. We wanted to integrate AI into this process, as a means to automatically and quickly document code.
What it does
docuMint automatically generates clear, concise, and comprehensive Markdown documentation for your code, all with the power of AI. It supports single file uploads and GitHub repository linking, providing an easy and efficient way to document various programming projects without being tied to a specific programming language. Through GitHub's authentication, users can view and modify their repositories through our service by generating and pushing documentation files.
How we built it
We built docuMint using a robust tech stack that includes Next.js, Tailwind CSS, and React. This combination allowed us to create a user-friendly interface and a responsive, efficient back-end. Our system leverages the power of these technologies to analyze code and generate documentation seamlessly with OpenAI. In addition, we utilize the GitHub API to give users the ability to modify and interact with their public and private repositories through our service, letting users pick folders within repositories to automatically generate documentation using AI.
Challenges we ran into
One of the major challenges was ensuring the accuracy and relevance of the generated documentation. No one on our team had prior experience with the OpenAI API or the GitHub API, so we enjoyed the challenges which arose with learning with those. We also faced hurdles in creating a system that could handle a wide range of programming languages and code complexities. Other challenges were related to Next.JS 14 which emphasizes a new routing concept than traditional page based routing.
Accomplishments that we're proud of
We are particularly proud of developing a tool that significantly reduces the workload of developers. Our system's ability to accurately interpret and document code of various complexities stands as a testament to our partnership's hard work and innovation. We are also proud of the perseverance we had through the many roadblocks which came up throughout the event.
What we learned
This journey taught us a great deal about natural language processing and the nuances of different programming languages. We also learned valuable lessons in user experience design, ensuring our tool is accessible and easy to use for everyone. Lastly, we learned about the GitHub and OpenAI APIs, both of which being essential tools for any software engineer and which will undoubtedly help us in our future classes and careers.
What's next for docuMint
Looking forward, we plan to smooth out the edges to make sure its a stable app that's available for the public to use. We're also exploring integrations with tools like MkDocs to further increase the capabilities of docuMint. To further existing features, we'd like to develop the GitHub integrations to include GitHub Actions and CI/CD pipelines, to allow our users to set up automated pipelines to create documentation easier.
Log in or sign up for Devpost to join the conversation.