Refactoring Executor

Refactoring Executor

Overview

Safely execute refactoring based on discovery reports using test-first verification and incremental steps. This skill automates/guides the refactoring process with continuous validation, metric tracking, and rollback capabilities to ensure zero regression.

When to Use This Skill

Invoke this skill in the following scenarios:

1. Post-Discovery Refactoring - After generating a refactoring-discovery report, execute recommended changes
2. Test-First Code Improvement - When refactoring requires validation at each step
3. Safe Incremental Refactoring - For complex refactoring that needs small, verifiable steps
4. Metric-Tracked Improvements - When you need before/after metrics for refactoring effectiveness

Prerequisites

Required Tools

• TypeScript: tsc, ts-node
• Testing: vitest (or jest)
• Linting: eslint
• Optional: ts-morph (for automated AST-based refactoring)

Input Requirements

• Refactoring discovery report (markdown format)
• All tests passing (green baseline)
• Clean git working directory (for rollback capability)

Core Capabilities

1. Report-Driven Execution

Parse refactoring discovery reports and execute recommended changes:

• Select target issue by priority
• Generate execution plan with step-by-step actions
• Validate feasibility and estimate effort

2. Test-First Verification

Ensure safety through continuous testing:

• Verify baseline (all tests pass before starting)
• Add characterization tests for missing coverage
• Run tests after each incremental change
• Validate types, lints, and performance metrics

3. Incremental Safe Steps

Support major refactoring patterns:

• Extract Method/Function
• Extract Class/Module
• Move Method/Field
• Rename Symbol
• Introduce Parameter Object
• Replace Conditional with Polymorphism
• Encapsulate Field/Collection
• Split Phase

4. Execution Logging

Track all changes with detailed logs:

• Before/after metrics (test time, bundle size, warnings)
• Each step's validation results
• Rollback points and procedures
• Decision rationale for each change

Safety Principles

Never Break Tests

• All tests must stay green throughout refactoring
• If a test fails, immediately revert the last change
• Investigate failure cause before proceeding

Small Steps, Frequent Validation

• One intent per change (rename → update usages → delete dead code)
• Commit after each successful validation
• Keep diffs reviewable and meaningful

Characterization Tests First

• Add tests to lock down existing behavior before structural changes
• Cover edge cases and invariants
• Use snapshot tests for complex outputs

Preserve Contracts

• Maintain existing API contracts
• Keep non-functional requirements (performance, security)
• Document any intentional breaking changes

Type-Driven Safety

• Use TypeScript's type system as a safety net
• Enable strict mode (strictNullChecks, noImplicitAny)
• Verify no type errors after each step

Workflow

6-Stage Execution Process

Stage 1: Parse Report and Select Target

Inputs: Refactoring discovery report (markdown)

Actions:

1. Parse report to extract issues, priorities, and recommendations
2. Ask user to select target issue (or auto-select highest priority)
3. Analyze dependencies and affected files
4. Verify no scope conflicts or circular dependencies

Outputs:

• Selected issue summary
• Affected file list
• Dependency graph

Safety Checkpoint:

• Issue is actionable and well-defined
• No conflicts with ongoing work
• Dependencies are manageable

Stage 2: Pre-Refactoring Validation

Actions:

1. Run full test suite and verify all pass
2. Run type checker (tsc --noEmit)
3. Run linter (eslint)
4. Capture baseline metrics:
   • Test execution time
   • Test count
   • Type error count (should be 0)
   • Lint warning count
   • Bundle size (if applicable)
   • Performance benchmarks (if available)

Outputs:

• execution-plan.md (from assets/execution-plan-template.md)
• Baseline metrics JSON

Safety Checkpoint:

• All tests pass (100% green)
• No type errors
• Git working directory is clean
• Baseline metrics captured

Rollback: If validation fails, stop and report issues to user

Stage 3: Add Characterization Tests

Actions:

1. Identify code paths lacking test coverage
2. Add characterization tests using assets/characterization-test-template.md:
   • Given: Initial state
   • When: Action/behavior
   • Then: Expected output (capture current behavior)
3. Run new tests to verify they pass
4. Commit characterization tests separately

Outputs:

• New test files (e.g., *.characterization.test.ts)
• Test coverage report

Safety Checkpoint:

• New tests pass with current implementation
• Tests cover critical paths affected by refactoring
• Tests are deterministic (no flakiness)

Rollback: Characterization tests are additive; if they fail, it indicates existing bugs

Stage 4: Execute Incremental Refactoring

Actions:

1. Break refactoring into smallest possible steps
2. For each step: a. Apply one refactoring pattern (see references/refactoring-patterns.md) b. Run verification (see Stage 5) c. If verification fails, rollback and adjust d. If verification passes, commit with descriptive message e. Log step results in execution-log.md

Supported Patterns (detailed in references/refactoring-patterns.md):

• Extract Method: Extract code block into named function
• Extract Class: Split class into focused classes
• Move Method: Relocate method to appropriate class
• Rename Symbol: Rename for clarity (auto-update references)
• Introduce Parameter Object: Group related parameters
• Replace Conditional with Polymorphism: Convert switch/if-else to strategy pattern
• Encapsulate Field: Add getters/setters for direct field access
• Split Phase: Separate computation from side effects

Outputs:

• Modified source files
• Git commits (one per step)
• Execution log entries

Safety Checkpoint (per step):

• Tests pass (no regressions)
• Types are valid (no new type errors)
• Linter passes (or warnings documented)
• Behavior unchanged (same outputs)

Rollback (per step):

git reset --hard HEAD~1  # Revert last commit
# Or use git stash if uncommitted

Stage 5: Step Verification

Actions (after each incremental change):

1. Run affected tests:

   pnpm vitest --run --reporter=json --filter <scope>
   

2. Run type checker:

   pnpm tsc --noEmit
   

3. Run linter:

   pnpm eslint --format=json <affected-files>
   

4. Optional: Run performance benchmarks
5. Capture metrics and compare to baseline
6. Log results to execution-log.md

Tool Integration (see references/tooling-playbook.md):

• scripts/verify-step.sh: Automated verification script
• scripts/ts-morph-refactors.ts: AST-based refactoring helpers

Outputs:

• Test results (JSON)
• Type checker output
• Lint results
• Step metrics

Safety Checkpoint:

• All tests pass (same or better than baseline)
• No new type errors
• Lint warnings same or fewer
• Performance within acceptable bounds (<5% regression)

Rollback: If any check fails, revert the step and document failure cause

Stage 6: Post-Refactoring Report

Actions:

1. Run full test suite and capture final metrics
2. Compare final vs. baseline metrics
3. Generate execution log using assets/execution-log-template.md:
   • Summary of changes
   • Steps taken with validation results
   • Before/after metrics comparison
   • Remaining work (if any)
   • Lessons learned
4. Update refactoring discovery report with completion status
5. Create pull request with:
   • Link to execution log
   • Metric improvements
   • Testing evidence

Outputs:

• execution-log.md (completed)
• Metrics comparison table
• Pull request description

Safety Checkpoint:

• All tests pass
• Metrics improved or maintained
• No regressions introduced
• Documentation updated

Example Usage Scenarios

Scenario 1: Extract Large Method

User: "Execute refactoring for ComparisonRunner.run() method (667 LOC)"

Process:

1. Parse: Load refactoring-discovery report, select ComparisonRunner issue
2. Validate: Run tests (✅), capture baseline (test time: 12.5s, CC: 45)
3. Characterization Tests: Add tests for pairing logic, summary generation, statistical analysis
4. Step 1: Extract pairing logic → PairingStrategy interface
   • Verify: Tests pass ✅, Types valid ✅, Lint clean ✅
   • Commit: "refactor: extract pairing logic to PairingStrategy interface"
5. Step 2: Extract summary generation → SummaryGenerator class
   • Verify: Tests pass ✅, Types valid ✅, Lint clean ✅
   • Commit: "refactor: extract summary generation to SummaryGenerator"
6. Step 3: Extract statistical analysis → StatisticalAnalyzer class
   • Verify: Tests pass ✅, Types valid ✅, Lint clean ✅
   • Commit: "refactor: extract statistical analysis to StatisticalAnalyzer"
7. Step 4: Refactor ComparisonRunner to use extracted components
   • Verify: Tests pass ✅, Types valid ✅, Lint clean ✅
   • Commit: "refactor: simplify ComparisonRunner using extracted components"
8. Report: Generate execution log
   • Before: 667 LOC, CC 45
   • After: 150 LOC (runner), 100 LOC (pairing), 150 LOC (summary), 100 LOC (analyzer)
   • CC reduced: 45 → 8 (runner), 6 (pairing), 8 (summary), 6 (analyzer)
   • Tests: 100% pass, no regressions
   • Time: 3.5 hours actual

Scenario 2: Split Module by Category

User: "Execute refactoring for metrics/index.ts (419 LOC, 3 categories)"

Process:

1. Parse: Load report, select metrics module issue
2. Validate: Baseline captured ✅
3. Characterization Tests: All metrics already well-tested ✅
4. Step 1: Create metrics/ranking.ts and move ranking metrics
   • Verify: Tests pass ✅
   • Commit: "refactor: extract ranking metrics to ranking.ts"
5. Step 2: Create metrics/latency.ts and move latency metrics
   • Verify: Tests pass ✅
   • Commit: "refactor: extract latency metrics to latency.ts"
6. Step 3: Create metrics/events.ts and move event metrics
   • Verify: Tests pass ✅
   • Commit: "refactor: extract event metrics to events.ts"
7. Step 4: Update metrics/index.ts to barrel export
   • Verify: Tests pass ✅, no breaking changes ✅
   • Commit: "refactor: convert metrics/index.ts to barrel export"
8. Report: Execution log shows clean split, backward compatible

Reference Materials

Refactoring Patterns (references/refactoring-patterns.md)

Detailed step-by-step guides for each supported pattern:

• Prerequisites and applicability conditions
• Incremental transformation steps
• Type safety checklist
• Rollback procedures
• Common pitfalls and solutions

Load this file when executing specific refactoring patterns.

Safety Checkpoints (references/safety-checkpoints.md)

Critical checkpoints throughout the refactoring process:

• Pre-refactoring validation checklist
• Per-step verification requirements
• Rollback triggers and procedures
• Metric threshold definitions

Load this file to ensure all safety gates are enforced.

Tooling Playbook (references/tooling-playbook.md)

Tool integration and usage examples:

• ts-morph scripts for AST-based refactoring
• vitest/jest command examples and failure diagnostics
• eslint integration with JSON output parsing
• tsc type checking with targeted file validation

Load this file when using automated tooling for refactoring.

Best Practices

DO:

• ✅ Always start with green tests
• ✅ Add characterization tests before structural changes
• ✅ Make smallest possible changes
• ✅ Commit after each successful validation
• ✅ Capture metrics before and after
• ✅ Document decisions in execution log
• ✅ Use TypeScript's type system as safety net
• ✅ Rollback immediately on failure

DON'T:

• ❌ Skip test validation
• ❌ Make multiple changes before verifying
• ❌ Ignore type errors
• ❌ Refactor without characterization tests
• ❌ Proceed with failing tests
• ❌ Forget to capture baseline metrics
• ❌ Skip documentation of changes

Integration with Development Workflow

With Refactoring Discovery

1. Run refactoring-discovery to generate report
2. Review report and prioritize issues
3. Use refactoring-executor to implement top-priority changes
4. Re-run refactoring-discovery to verify improvements

With Pull Requests

• Include execution log in PR description
• Link to refactoring discovery report
• Show before/after metrics
• Highlight test coverage improvements

With CI/CD

• Run verification script in CI pipeline
• Track metrics over time
• Fail builds on metric regressions
• Generate trend reports

Resources Summary

References (Load into context as needed)

• references/refactoring-patterns.md - Step-by-step guides for each pattern
• references/safety-checkpoints.md - Critical validation checkpoints
• references/tooling-playbook.md - Tool integration examples

Assets (Use in execution)

• assets/execution-plan-template.md - Plan template with steps and metrics
• assets/execution-log-template.md - Log template for tracking execution
• assets/characterization-test-template.md - Template for adding behavior tests

Scripts (Automation helpers)

• scripts/parse-report.ts - Parse refactoring discovery reports
• scripts/ts-morph-refactors.ts - AST-based refactoring utilities
• scripts/verify-step.sh - Automated step verification

These resources ensure safe, documented, and verifiable refactoring execution.