code-review

Reviews code diffs and pull requests against specs and standards, and guides responses to review feedback.

22|Updated Sep 10, 2026
One-click install
npx skills add https://github.com/Lynricsy/HyperSkills --skill code-review-lynricsy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/Lynricsy/HyperSkills/tree/main/skills/code-review
Command: npx skills add https://github.com/Lynricsy/HyperSkills --skill code-review-lynricsy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Reviewing a change well requires pinning the right base commit, separating spec compliance from code standards, verifying every finding, and ending with a clear verdict — steps that are easy to skip under time pressure. On the receiving side, acting on review feedback without verifying it first produces wasted work and broken code. ## Core Features & Use Cases - Structured diff review: Pins a base ref with git rev-parse, takes a three-dot diff, reviews tests first, and reports Spec and Standards findings in separate severity-labelled sections ending with an explicit merge verdict. - Security-aware delta review: Escalates on high-risk triggers (removed guards, auth changes, value transfer) with git blame archaeology, caller-counted blast radius, coverage-gap severity elevation, and concrete attack scenarios written to a report file. - Responding to review feedback: Enforces clarify-before-implementing on unclear items, evidence-backed pushback, YAGNI checks via caller greps, and per-item tested implementation in blocking-then-trivial-then-structural order. - Use Case: A reviewer flags that a guard clause is unreachable and asks you to delete it. Instead of complying, you verify against the code, find the guard is reachable, and reply in the PR thread with the evidence and a full-SHA permalink. ## Quick Start Ask the agent to review the diff on your current branch against origin/main and report findings with severities and a merge verdict.

Frequently Asked Questions about code-review

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

FAQPage Schema
How do I review a pull request diff with an AI agent?

Point the agent at the branch and base ref. It validates the base with git rev-parse, takes a three-dot diff, reviews the tests first, then reports Spec and Standards findings separately with severities and a Ready-to-merge verdict.

How should I respond to code review feedback before implementing it?

Read the whole review first, then verify each item against the code. If any item is unclear, ask all questions in one numbered round with recommended answers before editing anything, since answers can change what the clear items should do.

When does a code review need a security-focused pass?

A security pass triggers on high-risk changes: authentication, cryptography, external calls, value transfer, or removed validation. It adds git blame on removed guards, caller-counted blast radius, coverage-gap severity elevation, and one concrete attack scenario per finding.

Can this review a whole codebase for vulnerabilities?

No. It reviews a bounded delta — a diff, commit range, branch, or pull request — not a whole-codebase audit. Full codebase, feature, or threat-surface auditing belongs to a separate security-review skill.

Why does the review keep spec findings separate from standards findings?

Code can honour every standard while implementing the wrong thing, so merging the two axes lets one mask the other. Findings are gathered and reported in separate sections and never reranked across axes.

What tools does the review workflow require?

Only git is required; every command is plain git such as rev-parse, diff, log, and blame. The gh CLI is optional and only needed for replying to pull-request review threads.