Review and clean code comments

When to use this skill

Choose the scope. Default to diff-only cleanup: inspect the complete staged and unstaged patch and edit only the comments that patch introduces or changes. Use repository-wide cleanup only when the user explicitly asks for it; otherwise unchanged comments in a file are out of scope.
Apply the comment rules from the comment rules reference. In one line: comments are sparse, above the code, never narrated as edit history, never end-of-line — except Doxygen ///< trailing comments, the codebase convention for documenting public API (see the code review checklist).
Remove comment clutter: commented-out code; comments that merely restate the code or the method name; edit-history narration.
Preserve necessary comments: TODO, FIXME and similar work markers; linter, formatter, compiler, coverage, and generated-code directives; non-obvious reasoning, constraints, and business rules; pre-existing comments during diff-only cleanup.
Focus only on comments. Do not change executable code, behavior, or formatting unrelated to the moved or removed comment. Regenerate generated files instead of editing them.
Report the outcome with a before/after table (see examples). If the code already complies, say so — do not force changes onto compliant code.

Comment rules reference — the canonical merged rule set, with the source of each rule.
Before/after examples — worked examples built for this skill.
comment-cleanup skill — by Craig Motlin (Apache-2.0); the scope, remove, preserve, and placement rules are adapted from it.
clean-code-comments skill — on mcpmarket.com; the Comment Guidelines list is adapted from that page (transcribed by the ORE Studio user; the page declares no license).
Code review checklist — the Doxygen-on-public-API convention that defines the ///< exception.