Does it integrate with my existing CI/CD?

Yes, it uses the GitHub CLI to view PR checks and status rollups, ensuring that automated tests and builds are passing before recommending a merge.

How does the skill determine the review tier?

It calculates the total lines of code changed. Quick tier is for under 200 lines, Standard is 200-800 lines, and Deep tier is triggered for over 800 lines or any changes to security-sensitive files like auth or crypto modules.

What kind of feedback does the skill provide?

The skill generates a structured report categorizing findings as Blockers (must fix), Warnings (should fix), or Suggestions (optional improvements), covering quality, tests, CI, and architecture.

Can it detect security vulnerabilities?

In the Deep tier, it specifically scans for hardcoded secrets (API keys, tokens), analyzes input validation to prevent injection, and reviews changes to trust boundaries.

Tiered PR Reviewer

Name: Tiered PR Reviewer
Author: lev-os

bylev-os

•

コラボレーションツール

Conducts automated, tiered pull request reviews focusing on code quality, test coverage, and security vulnerabilities.

This skill transforms Claude into a senior engineering reviewer that evaluates pull requests through a structured, three-tier framework. It automatically determines the necessary review depth based on change volume and file sensitivity—ranging from quick quality checks for small PRs to deep architectural and security audits for large-scale changes. By integrating GitHub CLI tools and cross-referencing domain-specific principles, it provides consistent, actionable feedback that catches blockers early while filtering out the noise already handled by standard linters.

主な機能

01Integrated security scanning for hardcoded secrets and auth boundary changes

02Structured reporting with severity-based findings (Blocker, Warning, Suggestion)

03Automated PR scope detection and tier classification (Quick, Standard, Deep)

04GitHub CLI integration for real-time status check and metadata retrieval

05Automated test coverage analysis and source-to-test ratio validation

061 GitHub stars

ユースケース

01Validating test coverage and CI status consistency for large architectural changes

02Automating initial code reviews to catch common defects before human intervention

03Performing deep security audits on sensitive authentication and cryptographic modules

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lev-os/agents pr-review

For use in Claude.ai and ChatGPT