Nudge is a Golang application that sends nudge messages to users with overdue pull requests. It also has an optional integration with Slack. This has been inspired by the paper titled "Nudge: Accelerating Overdue Pull Requests Toward Completion" available at https://dl.acm.org/doi/abs/10.1145/3544791.
Nudge is available to use on GitHub Marketplace. With Nudge, you can automate the communication process for overdue pull requests, allowing you to stay on top of your project and keep it moving forward.
- Automated nudge messages for overdue pull requests
- Optional integration with Slack
- Support for multiple languages
- Customizable nudge message
- Fork the repository.
- Clone the repository.
- Install the dependencies:
go get -u ./... - Build and run the application:
make dev-docker
Run tests
make run-testsCreate a release
make releaseRun Locally On Mac
make run-dev-backendDesign Overview The Nudge system consists of three main components: A machine learning-based effort estimation model that predicts the lifetime of a given pull request, an activity detection module to establish what the current state of the pull request is, and an actor determination module to identify who would be need to take action
- Prediction Model. As of today this has not been open sourced and will be be made available soon.
- Activity Detection The role of the activity detection module is to help the Nudge system understand if there has been any activity performed by the author or the reviewer of the pull request of late. This helps the Nudge system not send a notification, even though the lifetime of the pull request has exceeded its predicted lifetime. This module serves as a gatekeeper that gives the Nudge system a “go” or “no go” by observing various signals in the pull request environment.
- Actor Identification. The primary goal of this module is to determine the blocker of the change (the author or a reviewer) and engage them in the notification, by explicitly mentioning them. This module comes into action once the pull request meets the criteria set by the prediction module and the Activity Detection modules. Once the Nudge system is ready to send the notification, the Actor Identification module provides information to the Nudge notification system to direct the notification toward the change blocker.
Nudge Workflow. The three modules are combined with a notification system to form Nudge as shown in Figure above.
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.