reviewer-discipline

Defines shared scope, severity, evidence, and reporting rules for diff-based code reviews.

Updated May 26, 2026
One-click install
npx skills add https://github.com/robot-denny/ai-sketchlab --skill reviewer-discipline-robot-denny
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reviewer-discipline
Source: https://github.com/robot-denny/ai-sketchlab/tree/main/.agents/skills/reviewer-discipline
Command: npx skills add https://github.com/robot-denny/ai-sketchlab --skill reviewer-discipline-robot-denny

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Independent code reviews often produce inconsistent reports with different severity scales, scopes, and formats that cannot be merged into one coherent action plan. This Skill establishes a single shared contract so multiple reviewers produce compatible, mergeable findings. ## Core Features & Use Cases - Diff-only scope rule: Enforces reviewing only code present in the diff, with conditional findings marked by explicit IF qualifiers when context is insufficient. - Unified severity scale: Provides one Blocker/Major/Minor/Nit scale with definitions, plus rules for resolving overlapping findings between reviewer domains. - Evidence and report standard: Requires a file and line citation for every finding and defines a fixed report structure with Summary, Findings table, Findings Detail, and a Clean section. - Use Case: When running accessibility, code quality, and performance reviews on the same pull request, each reviewer follows this contract so the three reports merge into one ranked list without duplicated or conflicting findings. ## Quick Start Apply the reviewer-discipline skill to review this diff and report findings using the shared severity scale and report structure.

Frequently Asked Questions about reviewer-discipline

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

FAQPage Schema
How do I make multiple code review reports consistent?

Use a shared review contract that fixes the severity scale, evidence standard, and report structure for every reviewer. This Skill defines one Blocker/Major/Minor/Nit scale and a single report template so independent reviews merge into one ranked list.

What severity levels should a code review use?

This Skill defines four levels: Blocker for must-fix defects that block users or expose data, Major for significant degradation, Minor for quality issues with workarounds, and Nit for best-practice notes. Inventing extra levels like Critical or High breaks report merging.

Should a code review cover code outside the diff?

No. The scope rule restricts review to code explicitly present in the diff, and unchanged code is out of scope. When context is insufficient, findings must be reported as conditional concerns with an explicit IF qualifier rather than asserted.

How do I avoid duplicate findings across multiple reviewers?

Assign each overlapping rule to a single owning reviewer and keep the others silent in merged reports. A reviewer running alone still raises everything, since the deferral only applies when another reviewer covers the rule.

What evidence should a code review finding include?

Every finding must cite a file path and line or range, carry a concrete recommended fix, and quantify impact where possible. Where the diff lacks line numbers, cite surrounding context such as the specific block or element.