code-review

Reviews code changes with severity-labeled findings across correctness, security, performance, and testing.

Updated Apr 17, 2026
One-click install
npx skills add https://github.com/mistakenot/skills --skill code-review-mistakenot
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/mistakenot/skills/tree/main/plugins/planning-workflow/skills/code-review
Command: npx skills add https://github.com/mistakenot/skills --skill code-review-mistakenot

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Reviewing pull requests and diffs consistently is hard: reviewers miss security issues, skip tests, or give vague feedback. This Skill enforces a structured review process with explicit severity labels so every change gets checked against the same dimensions and produces an actionable summary. ## Core Features & Use Cases - Severity-labeled findings: Classifies every issue as blocking, important, nit, suggestion, or praise so authors know exactly what must be fixed before merge. - Multi-dimensional review: Checks correctness, simplicity, security, performance, error handling, testing, architecture, and whether every changed line traces to the stated goal. - Automated check integration: Runs the project's typechecker, linter, and tests, reporting failures as blocking issues. - Use Case: After finishing a feature branch, ask for a review of the diff against main and receive a verdict (Approve / Request Changes / Comment) with file-and-line-specific findings grouped by severity. ## Quick Start Review the changes on my current branch against main and give me a severity-labeled summary with 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 with structured severity labels?▼

Run the review against the PR diff using gh pr diff or git diff main...HEAD, then classify each finding as blocking, important, nit, suggestion, or praise. The output is a summary with a verdict and issues grouped by severity with file and line references.

What does an AI code review check for?▼

The review covers correctness, simplicity, security, performance, error handling, testing, architecture, and surgical precision. It flags speculative abstractions, unrelated refactors, missing test coverage, and any automated check failures as blocking issues.

Can this review changes on a branch instead of a pull request?▼

Yes, it works on any branch by diffing against main with git diff main...HEAD. For pull requests it uses gh pr diff with the PR number instead.

Does the code review run tests and linters automatically?▼

Yes, the process includes running whatever automated tooling the project has, including typecheck, linter, and at minimum the affected tests. Any failing automated check is reported as a blocking issue.

When should I not rely on an automated code review?▼

Automated review is less reliable for judging product intent, subtle business logic, and large architectural trade-offs that require organizational context. Use it as a consistent first pass, with human reviewers validating domain-specific decisions.