Inspiration
This project was inspired by my daily work.
Every day, we spend excessive time improving task descriptions in Jira: adding acceptance criteria, test scenarios, steps, and explanations that take up a lot of time which could be better used for other activities.
What it does
Hakaboost is a Google Chrome extension that allows Jira Cloud users to accelerate the creation of acceptance criteria and use cases/test scenarios in Gherkin using current AI models, including the local version of Gemini Nano integrated into Chrome. Thanks to this tool, Hakaboost allows teams to save time, create high-quality descriptions, and focus their resources on other tasks.
Furthermore, based on the AI-enhanced description, it automatically generates the necessary issues in Jira to satisfy the identified tests. Additionally, and optionally, if the client has XRAY (By blend), Hakaboost can automatically create tests according to the client's desired type (Manual or Cucumber).
How we built it
We built it using a basic skeleton of web plugins and added functionalities iteratively. For this, we collaborated with Kiro, asking him to generate code for the documentation, review logic, and provide suggestions on using the APIs for AI and interacting with Jira and XRAY. Additionally, we asked Kiro for help with internal code documentation, guide comments, and English translation (we are native Spanish speakers).
Challenges we ran into
One of the most significant challenges was implementing simple yet effective rules for using Google Chrome's local AI and making it easily available to clients. Additionally, by using Jira, Xray, OpenAI, Gemini, and Chrome, the various features offered by the plugin had to be extensively tested.
Accomplishments that we're proud of
Integrate Google Chrome's local AI into a web plugin, activate it, and use it without requiring the user to interact with any external options or settings. In tests with development teams, we have measured the impact of Hakaboost, demonstrating a reduction of nearly 70% in the time teams spend writing and documenting tasks in Jira.
What we learned
We learned to use Kiro in a simple way and found it to be much simpler than other options like IntelliJ+Code GPT. We also learned about using Chrome's local AI, its implementation and limitations (compared to other cloud-based AI), and how various applications interact within a single application. In this regard, testing and error validation had to consider all possible variables.
What's next for HakaBoost
Enable its use with Jira Server, Trello, and other task managers for development teams. Improve test creation, with the ability to integrate different testing lifecycle models (issue types, states, test types, etc.).
Log in or sign up for Devpost to join the conversation.