本文へ移動
cccskills
無料GitHub で公開

code-quality

Code quality: cleanup, linting, formatting, quality gates.

インストール方法を見る

含まれるファイル(10)

  • SKILL.md6.0 KB
  • references/biome-rules-reference.md11.1 KB
  • references/cleanup-report-template.md3.1 KB
  • references/cleanup-scan-commands.md4.1 KB
  • references/code-cleanup.md11.7 KB
  • references/pyproject-template.toml15.4 KB
  • references/python-report-template.md3.1 KB
  • references/python-tool-commands.md4.4 KB
  • references/ruff-rules-reference.md10.0 KB
  • scripts/run_quality_gate.py2.4 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Code Quality

Four modes. Select by request signal:

SignalMode
Dead code, stale TODOs, unused imports, technical debtCleanup
Lint Python/JS/TS, run ruff, run biome, format codeLinting
Python quality checks, ruff+pytest+mypy+banditPython Quality Gate
Multi-language quality gate, general "quality check"Universal Quality Gate

Default: Python-specific requests run Python Quality Gate. General requests run Universal Quality Gate.


Mode A: Cleanup

Scan for 9 categories of technical debt: stale TODOs, unused imports, dead code, missing type hints, deprecated functions, naming inconsistencies, high complexity, duplicate code, missing docstrings. Follow references/code-cleanup.md for the full 3-phase workflow (SCOPE, SCAN, REPORT).

Key principles:

  • Read CLAUDE.md first for project conventions
  • Scan only what was requested; ask for a target rather than scanning everything
  • Exclude vendor/, node_modules/, .venv/, build/, dist/, generated/, .git/
  • Prioritize findings by impact/effort: Quick Wins (auto-fixable, <5 min) > Important (bugs, security) > Polish (style)
  • Apply auto-fixes only with explicit user permission; check first, fix, review diff, rerun checks

Mode B: Linting

Check or fix Python with Ruff and JavaScript/TypeScript with Biome. Read project instructions, pyproject.toml, and biome.json first; preserve configured rules.

TaskPythonJavaScript/TypeScript
Checkruff check .npx @biomejs/biome check src/
Format checkruff format --check .Included in Biome check
Fix lintruff check --fix .npx @biomejs/biome check --write src/
Formatruff format .npx @biomejs/biome format --write src/

Use project environment (./venv/bin/ruff). If configured, use make lint or make lint-fix. Check first; apply fixes only within requested scope. Review git diff, undo incorrect fixes, rerun checks. Do not change rules or install new tooling to get a pass.

Common fixes: F401 = remove/use import. I001 = sort imports. E501 = shorten lines. Biome noVar = use let/const. useConst = mark unchanged bindings.


Mode C: Python Quality Gate

Run checks in order. Read repository configuration first; project commands and thresholds override defaults.

Detect: Find pyproject.toml, setup.py, .python-version. Check tool availability: ruff and pytest required; mypy and bandit optional unless project requires them.

Execute:

OrderCommandCondition
1ruff check . --output-format=groupedRequired
2ruff format --check .Required
3mypy . --ignore-missing-imports --show-error-codesIf available
4pytest -v --tb=short --cov=src --cov-report=term-missingCoverage only when configured
5bandit -r src/ -ll --format=screenIf available

Run all checks even after one fails. Do not skip tests to get a pass.

Assess: Prioritize syntax/undefined errors, failing tests, security findings before style. Report per-tool status with actionable file:line diagnostics.

Fix: When authorized, run ruff check . --fix and ruff format ., review diff, rerun checks.


Mode D: Universal Quality Gate

Auto-detect languages and run configured checks. Read repository instructions first.

python3 ~/.claude/skills/code-quality/universal-quality-gate/scripts/run_quality_gate.py

Options: --staged (pre-commit), --lang python (single language), --fix (apply fixes), -v (verbose).

Language detection uses hooks/lib/language_registry.json:

LanguageMarkersTools
Pythonpyproject.toml, requirements.txtruff, mypy, bandit
Gogo.modgofmt, golangci-lint, go vet
JavaScriptpackage.jsoneslint, biome
TypeScripttsconfig.jsontsc, eslint, biome
RustCargo.tomlclippy, cargo fmt
Shell*.shshellcheck

A gate pass means required tools passed; it does not prove tests, builds, or business behavior passed.


Deep References

Load on demand when the signal matches.

SignalReferenceContent
Full cleanup workflow (9 categories)references/code-cleanup.md3-phase cleanup: SCOPE, SCAN, REPORT
Language-specific scan commandsreferences/cleanup-scan-commands.mdPer-language commands and expected output
Cleanup report formattingreferences/cleanup-report-template.mdStructured report template
Ruff rules, F401/E711/B006references/ruff-rules-reference.mdRuff rule reference by category
Biome rules, ESLint migrationreferences/biome-rules-reference.mdBiome rule reference by category
Python quality tool commandsreferences/python-tool-commands.mdDetailed tool command options
Python quality reportreferences/python-report-template.mdStructured report template
pyproject.toml templatereferences/pyproject-template.tomlTool configuration template

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

accounts

無料

Manage multiple Claude Code accounts: add, list, check, launch, and install shell aliases for 10+ isolated CLAUDE_CONFIG_DIR profiles.

日本語の概要は準備中です。原文の説明を表示しています。

notque/vexjoy-agent4412026年10月11日 更新

Improve architecture across modules by deepening interfaces.

日本語の概要は準備中です。原文の説明を表示しています。

notque/vexjoy-agent4412026年10月11日 更新

Assessment: read-only inspection, codebase overview, value analysis, health checks, ADR consultation, decision analysis, multi-perspective critique.

日本語の概要は準備中です。原文の説明を表示しています。

notque/vexjoy-agent4412026年10月11日 更新

Background memory consolidation — overnight review, merge, and injection payload for memory files.

日本語の概要は準備中です。原文の説明を表示しています。

notque/vexjoy-agent4412026年10月11日 更新

Jev-driven browser automation: Jev picks operations, programs execute, a text model writes field values only when Jev cannot pick one from the goal.

日本語の概要は準備中です。原文の説明を表示しています。

notque/vexjoy-agent4412026年10月11日 更新

Write, compose, integrate, and improve programs that call Jev, TypeSafe's System One judgment model.

日本語の概要は準備中です。原文の説明を表示しています。

notque/vexjoy-agent4412026年10月11日 更新

notque のスキルをすべて見る

このスキルの問題を報告する