Self-hosted alternatives to review-and-refactor

R

review-and-refactor

Automated code review

review-and-refactor is an agent skill from the github/awesome-copilot collection that reviews a project's code against its own coding guidelines and refactors it where needed. The agent takes the role of a senior engineer who has maintained a project for a long time. Before it reviews anything, it reads the coding guideline files in the .github/instructions folder and the .github/copilot-instructions.md file, so the standards it applies are the ones the project has written. It then reviews the code carefully and makes refactorings where they are needed, aiming for clean, maintainable code that follows those standards. Two constraints apply: it should not split up the code, keeping the existing files intact, and if the project has tests, they must still pass after the changes. It is useful for keeping a long-lived codebase tidy and consistent with its own rules, and for catching drift from the team's standards. Refactoring changes the code, so the changes should be reviewed as a diff and tested before they are merged. The skill's security scan reported a failure at the time of review, so the skill's contents should be read before it is installed. The skill is installed with the skills CLI from the github/awesome-copilot repository. Its skills.sh page showed about 11.0K installs, with Socket and Snyk passing and Gen Agent Trust Hub reporting a failure at the time of the scan.