change-classification

To prevent out-of-context commits, this Skill evaluates git diffs and issue context to classify changed files, flagging unrelated directory edits and grouping in-context changes into atomic commits.

Atomic commit grouping in code review workflows is the process of organizing in-context file changes into logical units by type or directory. This classification improves review focus by separating relevant modifications from uncertain or out-of-context files.

To detect secrets and large binaries in a git diff, the classification process analyzes filename and pattern signals alongside change size heuristics. It enforces warning or blocking actions when critical patterns like secrets or large additions are found.

Yes, you can flag first-touch files during a pre-PR check by evaluating filename and pattern signals. The classification identifies these new additions and highlights them in a classification table to ensure they receive extra review focus.

Yes, this code classification approach works with lockfile and auto-generated file changes by using secondary signals like sibling changes to evaluate their relevance. It specifically detects these file patterns and classifies them appropriately to prevent unnecessary review overhead.

You should use automated file classification instead of manual code review when you need to quickly surface out-of-context modifications, detect secret patterns, and group atomic commits. It serves as a first-touch filter to improve review quality before staging or committing.