How does confidence-based filtering reduce review noise?

The skill assigns a 0-100 score to every identified issue; only findings with a confidence score of 80 or higher are reported, filtering out pedantic nitpicks and hypothetical edge cases.

When is the best time to run this code review skill?

It is best used after completing major features, before merging to your main branch, or after significant tasks in a multi-step workflow where thoroughness is required.

Can this skill automatically fix code issues?

Yes, it can auto-fix critical issues (90+ confidence) and important issues (80+) using a 'Code Architect' agent, followed by a verification step to ensure the fix works.

What specific areas of my code does the AI review?

It runs three parallel reviews: Simplicity (DRY/elegance), Correctness (logic/edge cases), and Conventions (project patterns and naming).

Confidence-Based Code Review

Name: Confidence-Based Code Review
Author: jrc1883

byjrc1883

•

Collaboration Tools

Automates thorough code reviews using confidence-based filtering to eliminate noise and identify critical bugs, quality issues, and architectural inconsistencies.

The Confidence-Based Code Review skill streamlines the PR process by deploying parallel specialized agents to evaluate code for simplicity, correctness, and project conventions. By applying a strict 80+ confidence threshold, it filters out pedantic noise and false positives, ensuring developers only focus on meaningful improvements. The skill supports automated fixes for high-confidence issues (90+) and includes a verification step to ensure fixes are technically sound, making it an essential gatekeeper for merging production-grade code.

Key Features

015 GitHub stars

02Confidence-based filtering (80+ threshold) to eliminate false positives and reviewer noise.

03Integrated verification workflow that runs tests and re-reviews fixes automatically.

04Automated fixing capabilities for critical issues with 90+ confidence scores.

05Parallel multi-agent execution focusing on simplicity, bug detection, and project conventions.

06Flexible review scoping including staged changes, branch diffs, and specific commit ranges.

Use Cases

01Automated sanity checks after completing complex tasks in a multi-step subagent workflow.

02Technical debt cleanup by identifying and auto-fixing architectural pattern violations.

03Final quality gate before merging a major feature branch into the main repository.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jrc1883/popkit-claude pop-code-review

For use in Claude.ai and ChatGPT

Download Skill