Does this skill modify my git history?

No, Phased Review does not perform git operations like commits or branches; it applies fixes locally to your working directory for the user to review and commit.

What happens if a test fails during the review?

The workflow includes mandatory test gates. If a test fails at any stage, the process stops immediately until the failure is resolved to prevent regressions.

Where are the review findings stored?

Findings are recorded in a centralized, git-ignored markdown log file located in .claude/memory/reviews/ for easy traceability and historical reference.

Can I run only the security part of the review?

Yes, you can use the 'security' mode to skip general code reviews and focus exclusively on offensive, defensive, and architectural security analysis.

Is Phased Review limited to specific programming languages?

No, the skill is language-agnostic. it uses baseline detection to identify the specific test runners, linters, and type-checkers used in your project's ecosystem.

Phased Review Workflow

Name: Phased Review Workflow
Author: swannysec

byswannysec

0•

보안 및 테스팅

Conducts automated multi-stage code reviews and security audits with parallel sub-agents and gated test verification.

Phased Review is a comprehensive implementation review pipeline designed for Claude Code that orchestrates specialized sub-agents to analyze code quality, architecture, simplicity, documentation, and security. It follows a structured 11-stage sequence, enforcing mandatory test gates between every fix phase to ensure that autonomous improvements never break existing functionality. By blocking deep security audits until baseline stability is achieved and providing a centralized review log, this skill streamlines the path from development to production-ready code with high confidence.

주요 기능

01Centralized review logging in git-ignored markdown files for full traceability.

020 GitHub stars

03Multi-stage review pipeline covering code, architecture, simplicity, and documentation.

04Specialized security auditing using parallel Offensive, Defensive, and Architectural sub-agents.

05Mandatory test gates that halt the workflow if regressions or baseline failures are detected.

06Autonomous fix application for Critical, High, and Medium severity findings.

사용 사례

01Pre-release validation for feature branches and release candidates.

02Automated refactoring to eliminate over-engineering and improve documentation completeness.

03Deep-dive security audits and threat modeling for sensitive code paths.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add swannysec/robot-tools phased-review

For use in Claude.ai and ChatGPT

주요 기능

01Centralized review logging in git-ignored markdown files for full traceability.

020 GitHub stars

03Multi-stage review pipeline covering code, architecture, simplicity, and documentation.

04Specialized security auditing using parallel Offensive, Defensive, and Architectural sub-agents.

05Mandatory test gates that halt the workflow if regressions or baseline failures are detected.

06Autonomous fix application for Critical, High, and Medium severity findings.

사용 사례

01Pre-release validation for feature branches and release candidates.

02Automated refactoring to eliminate over-engineering and improve documentation completeness.

03Deep-dive security audits and threat modeling for sensitive code paths.