What is the 1-5 complexity scoring system?

It ranks tasks from Trivial (Level 1) to Very Complex (Level 5) based on affected files, lines of code, estimated time, and the number of technical unknowns involved.

How does the Quality Gates skill prevent over-engineering?

It utilizes a YAGNI gate that calculates a justified complexity ratio. If a solution is significantly more complex than the project tier warrants, the skill blocks the task and suggests simpler alternatives.

Can Quality Gates be used for code reviews?

Yes, it includes a Best Practices Pattern Library and Review Checklists to proactively detect anti-patterns and ensure code meets specific project standards.

When does the skill trigger an escalation to the user?

Escalation occurs when a task hits a blocking threshold, such as having more than three unanswered critical questions, missing dependencies, or failing after three consecutive attempts.

Quality Gates & Complexity Guardrails

Name: Quality Gates & Complexity Guardrails
Author: yonatangross

byyonatangross

•

生産性とワークフロー

Enforces development standards and prevents over-engineering by assessing task complexity and managing automated escalation workflows.

The Quality Gates skill provides a robust framework for Claude to evaluate task difficulty and maintain high architectural standards. It implements a 1-5 complexity scoring system, enforces YAGNI (You Ain't Gonna Need It) principles through justified complexity ratios, and monitors for 'stuck' patterns. By automatically blocking tasks with excessive unknowns or repeated failures, it ensures that AI development remains efficient, well-documented, and aligned with project-specific architectural tiers, ultimately preventing wasted compute cycles on poorly defined requirements.

主な機能

01YAGNI Gate Enforcement that blocks over-engineered solutions based on project tier.

0292 GitHub stars

031-5 Scale Complexity Scoring to size tasks by files, lines, and dependencies.

04Pre-task Gate Checks to identify missing requirements before implementation begins.

05Automated Escalation Protocols triggered after three failed attempts or critical blockers.

06LLM-as-Judge Validation for assessing depth, coherence, and technical accuracy.

ユースケース

01Reviewing proposed code against a pattern library to ensure best practices are followed consistently.

02Evaluating the complexity of a new feature to decide if it needs decomposition into subtasks.

03Detecting and stopping infinite retry loops when Claude is stuck on a specific bug or implementation.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit quality-gates

For use in Claude.ai and ChatGPT

Download Skill