菜单
Execute complete QA pipeline including code formatting with ruff, linting, import sorting with isort, type checking with mypy, and testing with pytest. Use when preparing code for commits, running pre-commit checks, ensuring code quality standards, or validating changes before pushing.
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
Validate Clean Architecture compliance including dependency flow verification (infra → interface → app → domain), module organization analysis, layer separation checks, and circular dependency detection. Use when reviewing code structure, analyzing imports, ensuring architectural principles, or validating refactoring changes.
Execute performance benchmarks for DataLoader configurations, training performance analysis, GPU utilization monitoring, and optimization validation. Use when analyzing performance bottlenecks, finding optimal training settings, validating speed improvements, or testing array bundling efficiency.
Validate that benchmark-training command stays in sync with learn-model command. Checks for missing CLI options, verifies test exclusion list accuracy, and identifies options that need to be added to benchmark-training when learn-model features are added. Use after adding new options to learn-model, before creating PRs that modify CLI commands, or when reviewing benchmark-training compatibility.
Execute cloud provider integration tests for Google Cloud Platform (GCP) and Amazon Web Services (AWS), validate S3 and GCS operations, test cloud storage authentication, verify bucket access, and debug cloud integration issues. Use when testing cloud storage features, validating parallel upload/download operations, or troubleshooting cloud connectivity.
Validate data pipeline configuration including array_type parameter verification, HCPE and preprocessing data format validation, storage configuration checks, and schema compliance verification. Use when configuring data sources, validating pipeline setup, debugging data loading issues, or ensuring data type consistency.
Manage Python dependencies using uv exclusively, add new packages, update existing dependencies, remove unused packages, and validate dependency compatibility. Use when adding packages, updating dependencies, resolving conflicts, or managing development dependencies. NEVER use pip directly.
| name | qa-pipeline-automation |
| description | Execute complete QA pipeline including code formatting with ruff, linting, import sorting with isort, type checking with mypy, and testing with pytest. Use when preparing code for commits, running pre-commit checks, ensuring code quality standards, or validating changes before pushing. |
Executes the complete quality assurance workflow for the Maou project following uv and Clean Architecture standards.
Run the complete QA pipeline:
uv run ruff format src/ && uv run ruff check src/ --fix && uv run isort src/ && uv run mypy src/ && uv run pytest
Execute these steps sequentially:
Normalize code style to 88-character line limit:
uv run ruff format src/
Fix linting issues automatically:
uv run ruff check src/ --fix
Organize imports with Black profile:
uv run isort src/
Validate type hints (REQUIRED before commits):
uv run mypy src/
Critical: Type hints are required for ALL functions, methods, and class attributes. No exceptions.
Execute test suite:
uv run pytest
For coverage analysis:
uv run pytest --cov=src/maou
All of the following must pass:
Type Safety: All functions must have type hints Code Style: 88-character line limit enforced Clean Architecture: Dependency flow maintained (infra → interface → app → domain) Test Coverage: New features and bug fixes have tests Documentation: Public APIs have docstrings
This pipeline matches the CI/CD checks that run on GitHub. Running it locally prevents CI failures.
If errors occur, follow this order:
uv run ruff format src/uv run ruff check src/ --fixuv run isort src/uv run mypy src/uv run pytest --tb=shortThis skill replicates the pre-commit hooks configured in .pre-commit-config.yaml. To install hooks:
uv run bash scripts/pre-commit.sh
See CLAUDE.md and AGENTS.md for complete development guidelines and code quality standards.