Purpose

Run comprehensive Quality Gate checks to verify code quality before committing. Uses parallel agent swarm execution for blazing-fast results. Non-destructive analysis only - reports issues without auto-fixing. Includes linting, formatting, type safety, tests, build verification, and security checks.

Variables

ENABLEJAVASCRIPT: true # Enable JavaScript/TypeScript quality checks ENABLEPYTHON: true # Enable Python quality checks ENABLESECURITYCHECK: true # Enable security vulnerability scanning ENABLEPARALLELEXECUTION: true # Use parallel agent swarm for faster execution SUPPORTEDPROJECTTYPES: javascript, typescript, python # Currently supported project types

Workflow

1. Parse User Request

- Identify quality check intent - User triggers: "run quality gate", "quality check", "check quality before commit", "verify code quality" - Example: "run quality gate" → Intent: comprehensive quality checks

1. Detect Project Type

- Check for indicator files: package.json (JS/TS), requirements.txt/pyproject.toml (Python) - Determines which cookbook workflow to use - Example: package.json found → JavaScript/TypeScript project

1. Route to Cookbook

- Based on detected type and ENABLE flags - JavaScript/TypeScript: IF package.json AND ENABLEJAVASCRIPT → javascript.md - Python: IF requirements.txt/pyproject.toml AND ENABLEPYTHON → python.md - Generic: IF no match → Run basic checks available in project - Example: TypeScript project + ENABLE_JAVASCRIPT=true → Route to cookbook/javascript.md

1. Execute Quality Checks

- IF: ENABLEPARALLELEXECUTION is true → Launch parallel agent swarm for all checks - Run all check phases defined in cookbook (linting, formatting, type checking, tests, build, security) - Tool: Task with runinbackground: true for each independent check - IMPORTANT: Non-destructive - only report issues, never auto-fix - Continue on failure - run all phases even if some fail (get complete picture) - Example: Launch 6 parallel agents (Linter, Formatter, TypeChecker, Tester, Builder, Security) → All complete in ~15s vs ~60s sequential

1. Generate Report

- IF parallel execution used → Collect all agent results using TaskOutput - Compile results from all phases - Include: specific file paths, line numbers when possible, error messages, actionable fix commands, performance comparison - Format: Clear sections per phase (✓ passed, ✗ failed), summary at end, execution time - Example: "Linting: ✗ 5 errors in src/utils.ts:23 - Run 'npm run lint:fix' | Performance: 75% faster (15s vs 60s)"

Cookbook

JavaScript/TypeScript Projects

• IF: The project has a package.json file AND ENABLE_JAVASCRIPT is true.
• THEN: Read and execute: .claude/skills/quality-gate/cookbook/javascript.md
• EXAMPLES:

- "run quality gate" - "quality check" - "check quality before commit" - "run all checks"

Python Projects

• IF: The project has requirements.txt or pyproject.toml AND ENABLE_PYTHON is true.
• THEN: Read and execute: .claude/skills/quality-gate/cookbook/python.md
• EXAMPLES:

- "run quality gate" - "quality check" - "verify code quality"

Generic Projects

• IF: No specific project type detected.
• THEN: Run basic checks available in the project and report.
• EXAMPLES:

- "run quality gate" - "check what we can"