AgentSkillsCN

refactor-phase-self-check

在完成实施工作后、提交前，或在准备接受评审时，将近期成果与已记录的模式、反模式以及原则相结合，进行综合评估与对比。

SKILL.md
--- frontmatter
name: refactor-phase-self-check
description: Evaluate recent work against documented patterns, anti-patterns, and principles with memory entity citations. Use after completing implementation, before commits, or when preparing for review.

Refactor Phase Self-Check Skill

Purpose

Systematic self-evaluation of recent work using memory-grounded principles. Orchestrates multiple sub-protocols to catch improvement opportunities BEFORE user review. Every evaluation cites specific memory entities.

When to Use

  • •After completing significant implementation
  • •Before declaring a task complete
  • •After refactoring existing code
  • •Before commits or PR submission
  • •When user asks for self-review
  • •When you want to verify your own work quality

Protocol Reference

For complete instructions, see: {{MEMORY_PATH}}/protocols/refactor_phase_self_check.md

The protocol contains:

  • •Complete 8-phase orchestration workflow
  • •Sub-protocols invoked (semantic_reflection, principle_check, anti_pattern_detection, pattern_check, code_smell_check)
  • •Memory entity citation requirements (every evaluation MUST cite specific entities)
  • •Evaluation criteria (readability, complexity, pythonic voice, design standards, performance)
  • •Test failure investigation rule
  • •Output format with citations and locations

Key Principle: Ground all evaluations in documented memory entities, not vague intuitions. Cite your sources. Every finding must reference a specific entity path.