Skip to content
Back to skills

React Change Review

ASecurity

Review a React or TypeScript UI diff for concrete behavior, accessibility, and performance risks. Use when a React review or hardening pass is requested.

  • 47 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added May 26, 2026
code-qualitytypescriptreactgitapiperformance

Works with

  • cli
  • api

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned September 22, 2026

npx -y skills add petekp/claude-code-setup --skill react-change-review --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of React Change Review?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for React Change Review
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/petekp-react-change-review/badge)](https://www.skillsdirectory.com/skills/petekp-react-change-review)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: react-change-review
description: "Review a React or TypeScript UI diff for concrete behavior, accessibility, and performance risks. Use when a React review or hardening pass is requested."
---

# React change review

Infer the review target from the request and Git state. Use the specified PR or range when given; otherwise inspect local changes and the branch's intended base. Read the diff and affected callers. Follow shared behavior beyond the changed files where it could break a consumer.

Load supporting guidance only for a relevant risk:

- `react-doctor`: running or interpreting React diagnostics. Prefer the repository's pinned command and reuse a current result for the same diff.
- `react-useeffect`: Effects, derived state, subscriptions, or cleanup.
- `vercel-composition-patterns`: component APIs or ownership changes.
- `vercel-react-best-practices`: suspected rendering, data-fetching, bundle, or server/client boundary issues.

Choose the canonical installed skill for each concern; do not load another vendor copy for the same guidance. Read individual rule references only when a suspected issue needs them.

Use [review checks](references/review-checks.md) for a substantial diff or an unfamiliar failure mode. Run required repository checks and additional verification that can resolve a concrete concern. A current successful check need not be repeated unless the diff or relevant environment changes.

Report evidence-backed findings in fix order. Each finding needs a failing condition, user or consumer impact, and a location. Merge findings with the same cause. Separate unresolved suspicions and verification gaps from confirmed defects. A clean review is a valid result.

Reviewing authorizes inspection and reporting. Fix only changes already authorized and within the immediate concern. Report unrelated diagnostics or cleanup as follow-ups; they do not expand scope.

Files in this skill

  • SKILL.md7.4 KB
  • agents/openai.yaml256 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.