code-review-standards

Apply a severity taxonomy and verdict protocol to software diffs.

15|5|Updated May 19, 2026
One-click install
npx skills add https://github.com/bobmatnyc/trusty-tools --skill code-review-standards-bobmatnyc
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: code-review-standards
Source: https://github.com/bobmatnyc/trusty-tools/tree/main/crates/trusty-code/src/assets/skills/code-review-standards
Command: npx skills add https://github.com/bobmatnyc/trusty-tools --skill code-review-standards-bobmatnyc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill eliminates subjective, inconsistent, or superficial code reviews by providing a rigorous, adversarial rubric that forces the AI to focus on production-critical issues rather than stylistic preferences.

Core Features & Use Cases

  • Severity Taxonomy: Standardizes findings into CRITICAL, HIGH, MEDIUM, and LOW categories to ensure focus on actual production risks.
  • Verdict Protocol: Implements a strict APPROVE/WARN/BLOCK logic to prevent code with critical defects from reaching production.
  • 80% Confidence Filter: Reduces noise by requiring the reviewer to have high confidence in a finding before flagging it, preventing speculative feedback.

Quick Start

Instruct the code-critic agent to apply the code-review-standards rubric to the current diff and provide a verdict.

Frequently Asked Questions about code-review-standards

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

FAQPage Schema
How do I standardize code review severity levels for production risks?

To standardize code review severity levels, categorize findings into CRITICAL, HIGH, MEDIUM, and LOW tiers. This severity taxonomy filters subjective style preferences and isolates actual production risks, ensuring reviewers focus only on high-integrity, actionable feedback.

What is an adversarial code review rubric and when do I need it?

An adversarial code review rubric is a structured protocol that forces rigorous, risk-focused software evaluation. You need it to eliminate inconsistent, superficial code reviews and prevent critical defects from reaching production through strict APPROVE, WARN, or BLOCK verdicts.

How do I apply a quality gate verdict protocol to a software diff?

Apply a verdict protocol to a software diff by evaluating the changes against a structured severity taxonomy, then issuing an APPROVE, WARN, or BLOCK verdict. This quality gate ensures diffs with critical defects fail automatically before deployment.

How can I reduce noise and speculative feedback during code reviews?

Reduce noise during code reviews by implementing an 80% confidence filter. This threshold requires reviewers to have high certainty in a finding before flagging it, preventing speculative feedback and ensuring only actionable production risks are reported.

Best way to automate code reviews without subjective style preferences?

The best way to automate objective code reviews is using a code-critic agent workflow with an adversarial rubric. This enforces a strict severity taxonomy and verdict logic, filtering out subjective style preferences to focus exclusively on production-critical risks.

What are the limitations of using a strict 80% confidence threshold for code analysis?

The limitation of an 80% confidence threshold for code analysis is that highly speculative but potentially critical edge-case vulnerabilities might be filtered out. This constraint reduces noise but requires strict adherence to ensure no uncertain yet severe production risks are missed.