sdlc

booch's avatarfrom booch

Software development lifecycle guidance. This skill should be used when planning development work, conducting code reviews, or establishing development workflows.

⭐17stars🔀3forks📁View on GitHub🕐Updated Jan 2, 2026
[Code Review]

When & Why to Use This Skill

This Claude skill provides comprehensive guidance on the Software Development Lifecycle (SDLC), specializing in a rigorous three-stage code review and refactoring process. It helps developers and AI agents maintain high code quality by enforcing standards for correctness, design, readability, and security throughout the development workflow.

Use Cases

  • Standardizing Development Workflows: Establish clear protocols for planning, commit frequency, and testing to ensure consistency across the engineering team.
  • Iterative Code Quality Assurance: Implement a sequential 3-reviewer process to identify edge cases, minimize coupling, and ensure code follows SOLID principles.
  • AI-Driven Refactoring: Leverage the skill to guide AI subagents in performing deep-dive code analysis followed by documented refactoring phases.
  • Best Practice Enforcement: Ensure every code change meets high standards for readability, simplicity, and security through structured evaluation criteria.
namesdlc
descriptionSoftware development lifecycle guidance. This skill should be used when planning development work, conducting code reviews, or establishing development workflows. Use proactively when reviewing code, preparing commits, creating pull requests, refactoring, or discussing development process and best practices. (user)

Software Development Lifecycle

This skill provides guidance on the software development lifecycle, with emphasis on code review and iterative improvement.

Code Review Process

All significant code changes require 3 sequential code reviews, each followed by refactoring.

Process

  1. Review Phase: Conduct a thorough code review, documenting findings
  2. Refactor Phase: Make changes based on the review findings
  3. Repeat: Pass the refactored code to the next reviewer

Each subsequent reviewer sees the refactored code from the previous cycle.

Review Criteria

Each review should evaluate:

  • Correctness: Does the code work as intended? Are edge cases handled?
  • Design: Does the code follow SOLID principles? Is coupling minimized?
  • Readability: Are names intention-revealing? Is the code easy to understand?
  • Simplicity: Does the code follow the Four Rules of Simple Design?
  • Testing: Are tests adequate? Do they serve as executable specifications?
  • Security: Are there any security vulnerabilities? (See the security skill for detailed guidance)

For AI Agents

When implementing the code review process with subagents:

  1. Complete the initial implementation
  2. Spawn a subagent with the role "Code Reviewer 1" to review and refactor
  3. Spawn a second subagent with the role "Code Reviewer 2" to review the refactored code and refactor again
  4. Spawn a third subagent with the role "Code Reviewer 3" for final review and refactoring

Each subagent should:

  • First produce a written code review documenting findings
  • Then make the refactoring changes based on those findings
  • Consider security implications in every review

Development Workflow

Before Writing Code

  • Understand the requirements fully before starting
  • Consider the design implications
  • Think about how the code will be tested

While Writing Code

  • Write tests first (TDD) when practical
  • Keep commits small and focused
  • Refactor continuously

After Writing Code

  • Run the full test suite
  • Conduct the 3-review process
  • Ensure documentation is updated if needed