refactor
Guided refactoring workflow for improving existing code without changing behavior. Ensures tests pass before and after changes.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Guided refactoring workflow for improving existing code without changing behavior. Ensures tests pass before and after changes.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Generate customized .claude/ harness framework with Fusion Architecture (GAN-inspired + Domain Specialists). TRIGGER when user wants to set up AI-assisted development structure, create a new project harness, add structure to existing projects, or mentions 'harness', 'scaffold', 'framework setup'. Supports all domains through template + dynamic generation.
Negotiates 'done' criteria between Generator, Evaluator, and architect-lead before implementation begins. Converts subjective goals into testable, gradable criteria with clear domain responsibilities.
Guided database migration workflow — from schema design to rollback documentation.
Production incident response workflow — triage, root cause analysis, and resolution.
Track, categorize, and prioritize technical debt across the codebase.
Save current progress with structured handoff artifact. Enables context reset for long-running tasks and session recovery.
| name | refactor |
| description | Guided refactoring workflow for improving existing code without changing behavior. Ensures tests pass before and after changes. |
| user-invocable | true |
You are facilitating the Refactoring process. Your role is to guide code improvement while preserving existing behavior.
Refactoring ensures:
IDENTIFY TARGET -> TEST BASELINE -> PLAN CHANGES -> INCREMENTAL REFACTOR -> VERIFY -> COMPLETE
1. Ensure tests pass (GREEN)
2. Make a small refactor
3. Run tests (should stay GREEN)
4. If tests fail (RED), revert and try smaller change
5. Repeat until refactor complete
IMPORTANT: If tests don't pass, fix them before refactoring.
Create refactoring plan with:
For each change:
# Refactor Plan: [Target]
## Objective
[What improvement is being made]
## Target Files
- [File 1]
- [File 2]
## Test Baseline
- Tests passing: [count]/[total]
- Coverage: [percentage]
## Changes Planned
### Change 1: [Name]
- **File**: [path]
- **Description**: [what change]
- **Risk**: Low/Medium/High
- **Depends on**: [previous change ID or "none"]
### Change 2: [Name]
- **File**: [path]
- **Description**: [what change]
- **Risk**: Low/Medium/High
- **Depends on**: Change 1
## Rollback Points
- [Commit ID]: [description of what was complete]
## Verification Steps
1. [Verification step 1]
2. [Verification step 2]
## Expected Outcome
[What the code will look like after]
## Status
DRAFT | TESTING | IN_PROGRESS | VERIFYING | COMPLETE
# Before
def process_order(order):
# 50 lines of mixed logic
# After
def process_order(order):
validate_order(order)
calculate_totals(order)
save_order(order)
# Before
class User:
# authentication methods
# profile methods
# notification methods
# After
class User:
# profile methods only
class UserAuth:
# authentication methods
class UserNotifications:
# notification methods
# Before
def calc(x, y):
return x * y + 10
# After
def calculate_adjusted_price(base_price, quantity):
return base_price * quantity + SERVICE_FEE
# Before
def validate_email(email):
if not email or '@' not in email:
return False
return True
def validate_phone(phone):
if not phone or len(phone) < 10:
return False
return True
# After
def validate_required(value, rules):
if not value:
return False
for rule in rules:
if not rule(value):
return False
return True
| Risk | Description | Approach |
|---|---|---|
| Low | Localized, simple change | Single commit |
| Medium | Multiple files, same patterns | Small commits, frequent tests |
| High | Complex changes, many files | Very small steps, checkpoints |
/refactor [target] [objective]
Example:
/refactor src/backend/services/user_service.py extract authentication logic