Can I choose the depth of the code review?

Yes, you can select 'BIG CHANGE' for an interactive section-by-section deep dive or 'SMALL CHANGE' for a condensed review with one question per section.

How are recommendations presented?

Every issue includes multiple numbered options (including 'do nothing') with specific ratings for implementation effort, risk, impact, and maintenance burden.

What is 'Plan Mode' in this Claude Code skill?

Plan Mode is an interactive stage where the skill evaluates your proposed code changes against engineering best practices before any actual code is modified.

What engineering principles does this skill prioritize?

It prioritizes 'engineered enough' solutions, aggressive DRY (Don't Repeat Yourself) practices, and non-negotiable, high-quality test coverage.

Interactive Code Review & Planning

Name: Interactive Code Review & Planning
Author: lev-os

bylev-os

0•

Collaboration Tools

Conducts rigorous, opinionated architectural and code quality reviews to ensure balanced engineering and robust test coverage.

This skill transforms Claude into a disciplined code reviewer that evaluates technical plans before implementation begins. It analyzes code through four critical lenses: architecture, quality, testing, and performance, while adhering to a philosophy of 'engineered enough'—avoiding both fragile hacks and over-engineered abstractions. By presenting concrete tradeoffs (effort, risk, impact) and requiring user consent for every recommendation, it ensures that every code change is intentional, DRY, and thoroughly tested.

Key Features

01Structured feedback loop that pauses for user approval before proceeding with changes.

020 GitHub stars

03Tradeoff-driven recommendations that include 'do nothing' options with risk assessments.

04Interactive workflow with 'Big Change' and 'Small Change' modes to control review depth.

05Multi-stage review covering architecture, code quality, testing, and performance metrics.

06Aggressive DRY enforcement and edge-case detection to prevent technical debt.

Use Cases

01Auditing legacy codebases for DRY violations and missing unit or integration test coverage.

02Reviewing complex architectural changes to ensure system boundaries and data flows are sound.

03Pre-implementation planning for performance-critical features to avoid N+1 queries and memory leaks.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lev-os/agents code-review

For use in Claude.ai and ChatGPT

Download Skill