capability-assess-code-quality

Invoke the assessment with an optional scope and path to measure cyclomatic complexity, function length, test coverage, duplication, and a maintainability index. The output is a structured report classifying each metric against thresholds with hotspots and recommendations.

The report covers cyclomatic and cognitive complexity, lines of code per file and function, line/function/branch test coverage, code duplication ratio, and a composite maintainability index scored 0-100. Each metric is classified against defined thresholds.

Yes, pass a path argument to restrict analysis to a specific file, directory, or package, and a scope argument to select metric groups like complexity or coverage. Omitting both assesses the full codebase.

No, the assessment is read-only. It inspects code and runs existing test commands for coverage data but never modifies files, making it safe to run at any point in the development workflow.

The assessment degrades gracefully by skipping test coverage metrics and noting 'Coverage: SKIPPED — no coverage tool detected' in the report. Other metric groups like complexity and size are still collected normally.

The skill is idempotent: it detects an existing quality report and checks staleness using git diff. If the codebase is unchanged, it confirms the existing report instead of re-assessing; if files changed, it proceeds with a fresh assessment.