speckit-review-code

Pull request code review involves checking changes against project guidelines and detecting bugs with high confidence. This skill targets working-tree changes, enforcing a confidence threshold of 80 to produce severity-grouped findings with file paths and actionable fixes.

Static analysis confidence scoring rates the certainty of detected code issues to filter false positives. This skill enforces a minimum confidence threshold of 80, outputting only severity-grouped findings for genuine bugs and guideline non-compliance with specific file paths and line numbers.

To enforce project-specific coding conventions, the skill checks changes against stored constitution or CLAUDE-style instructions within the `.specify/` structure. It validates adherence to explicit rules for style, naming, imports, and error handling, reporting violations with concrete fix suggestions.

Yes, this code review skill works with spec-kit project structures. It specifically targets projects containing the expected `.specify/` directory structure to review pull request and working-tree changes against stored guidelines like `constitution.md`.

The best way to catch functional and security bugs early is through high-confidence bug detection on pull request changes. This skill identifies actual logic errors, missing critical handling, and race or performance issues while filtering out likely false positives via an 80 confidence threshold.

Static analysis reports many false positives during issue triage because unfiltered tools flag every potential match. This skill reduces false positives by enforcing a confidence threshold of at least 80, ensuring only high-certainty bugs and guideline violations are reported with actionable fixes.