Code Quality Workflow for @j0kz/mcp-agents

Systematic approach to improving code quality using smart-reviewer and test-generator MCP tools.

When to Use This Workflow

• Before creating pull request - Pre-PR quality gate

• After significant refactoring - Verify no regressions

• During codebase audits - Systematic quality improvement

• When CI/CD quality checks fail - Fix issues systematically

• Adding new features - Ensure quality from start

5-Step Quality Pattern

1. Initial Assessment → Identify files needing review
2. Smart Review → Categorize issues (critical/moderate/minor)
3. Auto-Fix (Pareto 80/20) → Apply safe fixes automatically
4. Generate Tests → Cover untested code
5. Re-Review → Verify improvements

Quick Start

For Changed Files (Pre-PR)

Get changed files

git diff --name-only main...HEAD | grep -E '.(ts|js)$'

Batch Review

Tool: batch_review Input: { "filePaths": ["file1.ts", "file2.ts"], "config": { "severity": "strict" } }

Apply Auto-Fixes

Tool: apply_auto_fixes Input: { "filePath": "src/module.ts", "safeOnly": true // Always true for automation }

Generate Tests

Tool: write_test_file Input: { "sourceFile": "src/module.ts", "config": { "framework": "vitest", "coverage": 80 } }

Severity Configuration

Level Use For What It Flags

strict Production, APIs, Security ALL vulnerabilities, type violations, complexity >50

moderate Standard development, PRs Critical issues, complexity >70, major gaps

lenient Prototypes, experiments Only severe issues, breaking errors

For detailed severity configuration:

cat .claude/skills/code-quality-pipeline/references/severity-config-guide.md

Auto-Fix with Pareto Principle

Key Insight: 20% of fixes resolve 80% of issues

Safe Auto-Fixes (Apply Automatically)

• Formatting & indentation

• Import organization

• Unused code removal

• Simple type fixes

• Naming consistency

Manual Fixes (Review Required)

• Logic changes

• Refactoring suggestions

• Architecture improvements

• Complex type inference

For complete auto-fix patterns:

cat .claude/skills/code-quality-pipeline/references/auto-fix-patterns.md

Test Generation

Configuration

{ "framework": "vitest", // Standard for @j0kz "includeEdgeCases": true, // Boundary conditions "includeErrorCases": true, // Error paths "coverage": 80 // Target percentage }

What Gets Generated

• Unit tests for functions

• Edge cases (null, empty, boundaries)

• Error handling tests

• Async operation tests

• Mock setups

For test generation details:

cat .claude/skills/code-quality-pipeline/references/test-generation-guide.md

Common Patterns

Pattern 1: Fast Pre-Commit

Time: 1-2 minutes

1. Review staged files (moderate)
2. Apply safe auto-fixes
3. Run tests

Pattern 2: Comprehensive Pre-PR

Time: 5-15 minutes

1. Review all changes (strict)
2. Auto-fix safe issues
3. Manual fix critical issues
4. Generate missing tests
5. Verify coverage >75%

Pattern 3: Legacy Code Improvement

Time: 30-60 minutes

1. Review with moderate severity
2. Apply all safe fixes
3. Fix critical issues manually
4. Generate test suite
5. Re-review for verification

For complete workflow examples:

cat .claude/skills/code-quality-pipeline/references/complete-workflow-examples.md

Expected Outcomes

After Auto-Fix (Safe Only)

Issues: -73% average reduction Formatting: 100% consistent Imports: 100% organized Dead code: 100% removed

After Full Pipeline

Critical issues: 0 Complexity: <50 Coverage: >75% Maintainability: >80

Issue Priority Guide

Critical (Fix Before Merge)

• Security vulnerabilities

• Type safety violations

• Resource leaks

• Null/undefined errors

Moderate (Fix If Time)

• High complexity (>70)

• Missing documentation

• Performance issues

• Duplicate code

Minor (Future Cleanup)

• Style preferences

• Micro-optimizations

• Comment improvements

Integration with Other Tools

With orchestrator-mcp

Tool: run_workflow Input: { "workflow": "pre-merge", "params": { "files": ["..."] } }

With modular-refactoring

After quality pipeline, if complexity >50:

• Apply modular-refactoring-pattern

• Split files >300 LOC

• Extract to helpers/utils

Troubleshooting

Issue Solution

"Too many issues" Use auto-fix for 80% reduction

"Tests failing" Check if bugs found (good!) or mocks needed

"Auto-fix broke code" Restore from .backup/, use safeOnly=true

"Inconsistent results" Check severity level appropriateness

Quick Commands

After quality improvements

npm test # Run tests npm run test:coverage # Check coverage npm run update:test-count # Update badges git diff # Review changes

Related Skills

• mcp-workflow-composition: Orchestrate multiple tools

• modular-refactoring-pattern: Reduce complexity

• testing-patterns-vitest: Deep testing guidance

Scripts Available

Check the scripts/ directory for automation:

ls .claude/skills/code-quality-pipeline/scripts/

For project standards: .claude/skills/project-standardization/SKILL.md

For workflow orchestration: .claude/skills/mcp-workflow-composition/SKILL.md