What it does
Z.ai Code Review is a step you add to a GitHub Actions workflow that runs when a pull request is opened or updated. It sends the changes to a model from Z.ai and posts the result as a comment on the pull request, flagging possible bugs and ways to improve the code. You provide a Z.ai API key as a repository secret. Optional inputs let you pick the model (glm-4.7 by default), replace the reviewer's system prompt, rename the reviewer shown in the comment, exclude files by pattern (lock files are excluded by default) and cap how many characters of the diff are sent. The prompt can be rewritten to focus on a particular concern, such as security issues or a house coding standard. The project is MIT licensed and has a GitHub Marketplace listing.
Capabilities
- Runs automatically on opened and updated pull requests
- Posts bug findings and improvement suggestions as PR comments
- Configurable model, system prompt and reviewer name
- File exclude patterns and an optional cap on diff size
- MIT licensed
When to use it
- Getting an automated first-pass review on every pull request in a GitHub repository
- Reviewing with a custom prompt, for example one focused on security
- Teams that already hold a Z.ai API key