code-review

Review pull requests and diffs for correctness, security, and merge decisions.

1|Updated May 6, 2026
One-click install
npx skills add https://github.com/jacob-balslev/skill-graph --skill code-review-jacob-balslev
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/jacob-balslev/skill-graph/tree/main/marketplace/skills/code-review
Command: npx skills add https://github.com/jacob-balslev/skill-graph --skill code-review-jacob-balslev

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Code-review turns a pull request or diff into a structured check for correctness, clarity, security, performance, and alignment with project conventions, reducing the chance that regressions, vulnerabilities, or contract mistakes slip into production.

Core Features & Use Cases

  • Pre-review fact-gathering to align on scope, intent, and acceptance criteria before reading code.
  • Tests-first read order to judge whether the change is described and validated correctly before implementation details.
  • Severity-grading rubric and disciplined comments to ensure reviewers request the right level of change and communicate clearly.
  • AI-generated diff verification to prevent rubber-stamping by explicitly checking tests, error handling, and security assumptions.
  • Merge decision gate that forces an explicit outcome (approve / request changes / close without merge) based on blockers and change-requested items.

Quick Start

Use the code-review skill to review the diff for a pull request and determine whether to approve, request changes, or close without merge based on the rubric.

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 to catch security and performance issues before merging?

Review a pull request by analyzing the diff against a severity rubric for correctness, security, performance, and conventions. This structured process uses a tests-first reading order and call-site validation to produce an explicit merge decision.

What is the best way to review AI-generated code diffs before shipping?

Reviewing AI-generated code diffs requires heightened verification of tests, error handling, and security assumptions to prevent rubber-stamping. The review applies a six-phase workflow with a severity rubric to grade issues and determine an explicit merge outcome.

How do you grade issues when doing a code review on a pull request?

Grading issues during a code review involves applying a severity rubric to categorize correctness, clarity, and security problems. Disciplined comments are generated to request the appropriate level of change, culminating in an explicit approve, request changes, or close decision.

Can I use a structured code review process for both human-authored and AI-generated pull requests?

Yes, a structured code review process applies to both human-authored and AI-generated changes across varied pull request sizes. It aligns on scope and acceptance criteria first, then validates tests and implementation details to output a final merge decision.

When should I use a tests-first reading order for pull request reviews?

A tests-first reading order should be used during pull request reviews to judge whether the change is correctly described and validated before examining implementation details. This approach catches contract mistakes and regressions early in the review workflow.

Why does my code review process fail to catch regressions before merging?

Code reviews fail to catch regressions when they lack a structured severity rubric and pre-review fact-gathering. Implementing a six-phase workflow with call-site validation and an explicit merge decision gate prevents vulnerabilities and contract mistakes from slipping through.