Windsurf Skills let you define custom agent instructions and the code review use case is immediately useful

The Windsurf Skills feature introduction https://www.youtube.com/shorts/211abbXBOdo is the kind of capability that sounds like a power user feature until you see the code review use case demonstrated and realise it applies to almost every development team.

Creating a custom skill that defines exactly how you want code reviews structured, what issues to look for, how to format the summary, what severity categories to use, and saving it either for the current project or globally, changes the consistency of AI-assisted code review significantly.

The @ invocation from the prompt window to apply any saved skill is the right interaction model. You are not configuring a new prompt each time. You are applying a defined process configured once that applies consistently.

Multi-format context support for PDFs, text files and docx files is the integration with existing documentation that matters for organisations with established engineering standards. A skill that references your actual internal code standards document rather than generic best practices produces more relevant review output.

The potential uses beyond code review, release notes generation, refactoring guidance, hotfix management, are the starting points for a library of reusable engineering process automations the team builds over time.

For engineering teams with established code review standards: how much time do you spend on code review per PR and is the bottleneck the actual review or the documentation of findings?