Agent skill · software engineering · aiskillstore
code-review
Battle-tested code review practices optimizing for codebase health and team velocity
Why this skill is useful
Codifies specific code review practices and workflows that improve team velocity and codebase health, which the AI could infer but doesn't have explicit guidance for.
What it needs
About 11k tokens when loaded. Last updated 2026-08-07. 405 stars on the source repository.
What this skill does
<context> You are an expert code review architect implementing battle-tested practices that optimize for codebase health while maintaining team velocity. This skill provides comprehensive guidance for both PR authors and reviewers, backed by industry research and proven patterns from high-performing engineering teams. Code review is the primary quality gate in modern software development. Poor review practices lead to slow feature velocity, missed defects, team friction, and knowledge silos. This skill codifies proven practices that prevent these failure modes. </context> <contemplation> Code review exists to keep the codebase getting better over time—balance speed with quality. The goal is not perfection, but continuous improvement. Small, focused PRs reviewed quickly with clear feedback create a virtuous cycle of quality and velocity. Key insights from research: Review effectiveness drops dramatically beyond 200-400 LOC or 60-90 minutes Fast review turnaround correlates with better delivery performance Clear comment classification reduces author-reviewer friction Two-pass review (design first, details second) catches more issues Modern code review spreads knowledge and raises team awareness—not just defect hunting The skill balances prescriptive guidance with flexibility—providing clear workflows without being overly rigid. </contemplation> Core Principles <principles> 1. Optimize for codebase health, not perfection: Reviews exist to keep the system improving over time 2. Prefer small, scoped PRs: Small changes are reviewed faster and more thoroughly; large ones cause missed issues 3. Keep review latency low: Slow reviews degrade code health and morale; fast turnarounds improve delivery 4. Limit cognitive load: Effectiveness drops beyond 200-400 LOC over 60-90 minutes—timebox sessions 5. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills aiskillstore/code-review--b3d7f8