Does this skill produce a lot of false positives?

No. It includes a dedicated confidence-scoring phase where a secondary agent evaluates every finding, filtering out any issues that don't meet an 80% confidence threshold.

Is it possible to run only specific types of reviews?

Yes, you can provide optional arguments to focus the review on specific aspects like security, performance, or test coverage depending on your current needs.

Can I customize the review based on my project's standards?

Yes. The skill automatically searches for and respects instructions in CLAUDE.md, README.md, and other local documentation to ensure the review aligns with your specific coding standards.

How does this differ from a standard linter or compiler?

Unlike linters that focus on syntax and formatting, this skill uses specialized LLM agents to detect deep logic bugs, architectural flaws, and security vulnerabilities that static analysis tools often miss.

Does this skill require an active internet connection?

The skill requires model invocation to analyze the code changes, but it operates on your local git diffs and uncommitted files.

Local Change Code Reviewer

Name: Local Change Code Reviewer
Author: NeoLabHQ

byNeoLabHQ

•

542

•

Seguridad y Pruebas

Performs a multi-agent, comprehensive audit of uncommitted local changes to identify bugs, security risks, and quality improvements before you commit.

This skill automates the pre-commit review process by deploying specialized AI agents to analyze local diffs for security vulnerabilities, logic bugs, and architectural alignment. It provides a structured report including a Quality Gate assessment, confidence-scored issue lists, and actionable refactoring suggestions based on project-specific guidelines like CLAUDE.md. It is designed for developers who want to ensure high-quality, secure code while maintaining velocity by catching critical errors before they ever reach the repository or CI pipeline.

Características Principales

01Multi-agent parallel analysis focusing on security, bugs, quality, and contracts

02Seamless integration with project-specific rules in CLAUDE.md or README.md

03542 GitHub stars

04Detailed reporting with impact ratings and specific code improvement examples

05Automated confidence-scoring to filter out false positives and low-impact nitpicks

06Historical context review using git history to prevent regression errors

Casos de Uso

01Validating complex refactors before pushing to shared branches

02Ensuring new contributors adhere to project-specific architecture and style guides

03Automating security audits for local changes in sensitive codebases

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add neolabhq/context-engineering-kit review-local-changes

For use in Claude.ai and ChatGPT

Download Skill