piv-review-changes

Reviews recently changed git files for bugs, security issues, and standards compliance, then writes a report.

Updated Aug 28, 2026
One-click install
npx skills add https://github.com/az9713/claude-code-hooks-tutorial --skill piv-review-changes-az9713
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: piv-review-changes
Source: https://github.com/az9713/claude-code-hooks-tutorial/tree/main/skills/.claude/skills/piv-review-changes
Command: npx skills add https://github.com/az9713/claude-code-hooks-tutorial --skill piv-review-changes-az9713

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Catching logic errors, security vulnerabilities, and standards violations before code is committed is tedious and easy to skip under time pressure. This Skill automates a structured technical review of your uncommitted changes so issues are found before they reach the repository. ## Core Features & Use Cases - Git-aware change detection: Runs git status, git diff, and lists untracked files to identify exactly what changed, then reads each file in full for context. - Multi-category analysis: Checks for logic errors, security issues (SQL injection, XSS, exposed secrets), performance problems, code quality violations, and adherence to documented codebase standards. - Structured report output: Writes a severity-tagged report (critical/high/medium/low) with file paths, line numbers, and fix suggestions to .claude/code-reviews/. - Use Case: Before committing a feature branch, invoke the review to get a written report flagging a missing error handler on line 42 and an exposed API key, with concrete fix suggestions for each. ## Quick Start Review my current uncommitted changes for bugs and security issues and write a report before I commit.

Frequently Asked Questions about piv-review-changes

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review uncommitted git changes before committing?▼

Run the review on your working tree: it uses git status, git diff HEAD, and git ls-files to find modified and new files, reads each in full, and reports logic errors, security issues, and standards violations with line numbers and suggested fixes.

What does an automated code review check for?▼

It checks five categories: logic errors (off-by-one, race conditions), security issues (SQL injection, XSS, exposed secrets), performance problems (N+1 queries, memory leaks), code quality (DRY violations, poor naming), and adherence to documented codebase standards.

Where is the code review report saved?▼

The report is written as a Markdown file to the .claude/code-reviews/ directory. It includes change statistics and one entry per issue with severity, file path, line number, explanation, and a fix suggestion.

Can this review enforce project-specific coding standards?▼

Yes. The review first reads CLAUDE.md, README.md, and documentation in the docs directory to learn your project's linting, typing, logging, and testing standards, then flags deviations from those documented conventions.

When should I not rely on an automated pre-commit review?▼

It reviews only uncommitted changes in the working tree, so it will not catch issues in already-committed history or architectural problems spanning untouched files. Use it as a pre-commit gate, not a substitute for full codebase audits.