hmj1026
UserReusable Claude Code dev harness plugin — agents, commands, skills, hooks, opt-in stack modules
Categories
Indexed Skills (25)
claude-health
Claude Code config health check + plugin sync. Use when: auditing .claude/ structure, checking naming, verifying hook setup, detecting plugin version drift, syncing installed assets. Not for: skill quality (use skill-health-check), code review (use codex-code-review). Output: health report + fix recommendations.
code-explore
Pure Claude code investigation. Use when: tracing execution paths, understanding architecture, diagnosing issues. Not for: dual-perspective review (use code-investigate), code review (use codex-code-review). Output: analysis report with findings.
code-investigate
Dual-perspective code investigation. Use when: deep code analysis needing both Claude and Codex perspectives. Not for: quick exploration (use code-explore), code review (use codex-code-review). Output: integrated findings from dual analysis.
skill-health-check
Validate skill quality against routing, progressive loading, and verification criteria. Use when: auditing skills, checking skill health, reviewing skill design. Not for: code review (use codex-code-review) or doc review (use doc-review). Output: health report with per-skill ratings + Gate.
yii1-security-audit
Yii 1.1（PHP 5.6）框架安全白盒靜態審計。Use when：對 Yii 1.1 專案查找漏洞、驗證 accessRules/checkAccess/CSRF/SQL injection/CUploadedFile/Session 設定，或框架安全複查（即使用戶只說「檢查這個 Controller 有沒有安全問題」也應觸發）。Not for：Yii2 專案、非 PHP 專案、純業務邏輯演算法審查。Output：依 AUTH/CSRF/XSS/SQL/CFG/LOGIC/FILE 分類、含檔名:行號 + 可觀測 PoC + Yii API 修復建議的審計報告。
adaptive-dev-workflow
Workflow router for substantial changes. Use when: a feature, bugfix, refactor, security, perf, or OpenSpec change needs classifying into Feature Delivery, Bug Investigation & Fix, or Lightweight Maintenance before heavy context loads. Not for tiny edits, investigation already underway, code review, or apply-ready OpenSpec tasks. Output: workflow classification + required artifacts + gate checklist.
agent-architecture-audit
Full-stack diagnostic for agent and LLM applications — audits the 12-layer agent stack for wrapper regression, memory pollution, tool-discipline failures, hidden repair loops, and rendering corruption. Use when: shipping an agent/LLM feature, an agent degrades after adding wrapper/memory/tool layers, the same model works in the playground but breaks in your wrapper, or debugging agent behavior >15 min with no root cause. Not for: general code debugging, code review, or security scanning (use security-review). Output: severity-ranked findings with code-first fixes.
agy-commit
Delegate git smart commit to agy-cli. Use when: committing changes via agy-cli's git-smart-commit skill (offload commit batching to Antigravity). Not for: a single simple commit, or when agy-cli is unavailable. Output: grouped commits created by agy-cli plus a git log summary.
agy-fast-worker
Offload a self-contained, clearly-specified mechanical task to the agy (Antigravity) CLI backend and relay its result, with the heavy output quarantined in the agy-fast-worker subagent. Use when: a mechanical batch with a precise task spec (target files + change intent + verification command) should run on the cheap high-throughput agy tier (default model `Gemini 3.5 Flash (High)`) instead of the in-process sonnet fast-worker, and the agy CLI is available. Not for: ambiguous specs (escalate), work needing this conversation context (agy sees a fresh session — the prompt must be self-contained), or the default in-process path (use plain `fast-worker`). Output: the verification result + the working-tree-derived edited-file list, or an honest BLOCKED report — never simulated.
bug-fix
Bug fix workflow. Use when: fixing bugs, resolving issues, regression fixes. Not for: new features (use feature-dev), understanding code (use code-explore). Output: fix + regression test + review gate.
bug-investigation
Root-cause investigation workflow for unexpected behavior, test failures, performance regressions, data inconsistencies, and cross-layer data-flow tracing. Use when users ask to investigate or trace a bug, regression, flaky test, or root cause (including 測試失敗、效能異常、調查 Bug、追蹤資料流). Not for direct implementation when the root cause is already confirmed, ordinary code review, or pure feature work. Output: a phase-based evidence report, confirmed root cause or explicit blocked state, solution options, stop-loss state, and next actions.
legacy-code-characterization
Write characterization tests for untested legacy code to lock current behavior before refactoring. Use when users want to safely refactor legacy controllers/models/services, add regression safety nets, or prepare code for extraction. Trigger words: characterization test, lock behavior, legacy test, safe refactor, 行為鎖定, 特徵測試, 安全重構. Not for: greenfield code, already well-tested modules, or pure bug fixes. Output: characterization test files + coverage delta report.
multi-ai-sync
Compare and synchronize Claude-first agent configuration across Codex, Gemini, and Antigravity (.agent). Generate read-only plans, create OpenSpec tasks, apply safe sync changes with dry-run preview and writable-path fallback, and validate post-sync results. Use when aligning `.claude` skills, commands, hooks, agents, or orchestration rules to other AI platforms, when comparing multi-platform agent setup drift, or when preparing a Claude-to-Codex/Gemini/.agent migration. Also matches requests such as `sync claude to codex`, `align multi-platform AI config`, `同步 claude 技能到 codex`, or `對齊多 AI 設定`. Not for single-platform edits, reverse sync, or missing `.claude` source.
php-pro
Use when building, debugging, refactoring, testing, or reviewing PHP code in Laravel, Symfony, generic modern PHP, or legacy PHP 5.6 + Yii 1.1 codebases. Detect the active runtime first, then load only the matching reference set; for Laravel, confirm the resolved major version and load the version-specific reference before touching bootstrap/app.php, middleware, exceptions, auth, testing infrastructure, or Laravel-coupled packages. Not for frontend-only or non-PHP tasks. Output: runtime-matched implementation or review guidance with compatibility and verification gates.
php56-yii-dev
Generic backend development workflow for legacy Yii 1.x applications running on PHP 5.6 that must stay compatible with future PHP 7 upgrades. Use when implementing, refactoring, debugging, testing, or reviewing Yii 1.x backend code such as Controllers, FormModels, CActiveRecord, services, repositories, DAO queries, validation rules, security hardening, or domain-driven design boundaries. Includes pragmatic DDD placement, Context7-first knowledge refresh, and full TDD guidance. Not for frontend-only work, documentation-only edits, or non-PHP stacks. Output: a PHP 5.6-safe implementation or review plan with TDD and compatibility evidence.
fastapi-pro
FastAPI + SQLAlchemy 2.0 async patterns + review checklist: routers, Depends DI, Pydantic schemas, session/transaction discipline, Alembic safety, auth/CORS/errors. Use when writing or reviewing FastAPI routes, async repositories, or migrations. Not for non-web Python. Output: layered, transaction-safe API.
nextjs-15-5-notes
Next.js 15.5 (current stable; the 15.x line ends at v15.5.19 — 15.6 never shipped stable, canary only before the jump to 16.0) signature features and traps. Use when writing or reviewing code in a Next.js 15.5 project, or a package whose package.json constraint is ^15.5. Covers next typegen, stable typed routes, beta Turbopack production builds, and the next lint deprecation (removed in 16). Not for application business logic — load when working on framework-touching code (App Router files, next.config, build tooling) or planning a 15.5 -> 16 upgrade. Output: version-specific guidance, migration traps, and verification gates.
nextjs-16-notes
Next.js 16 (current stable major, now at 16.2.x) signature features and the breaking-change traps from 15.x -> 16.0. Use when writing or reviewing code in a Next.js 16 project, or planning an upgrade from Next.js 15.x. Covers Turbopack-by-default (dev and build), the async-only Request APIs (params/searchParams/cookies/headers), the Node.js 20.9+ floor, the removal of next lint and AMP, next/image priority -> preload, removed runtime-config APIs, and the 16.1 next upgrade CLI. Not for application business logic — load when working on framework-touching code or planning a 15.x -> 16 upgrade. Note: Next.js 16 supports React 18.2+ or 19 — React 19 is recommended but NOT required. Output: version-specific guidance, migration traps, and verification gates.
phpunit-10-notes
PHPUnit 10.x (February 2023) signature features and the breaking-change traps from 9 → 10. Use when writing or reviewing tests in a PHPUnit 10 project, or when migrating a PHPUnit 9 suite to 10. Covers PHP 8.1 floor, the attribute-style annotation system (#[DataProvider] / #[CoversClass] / #[Group] / #[Test]) that replaces (but does not yet remove) doc-comment annotations, the test runner CLI rewrite, TestListener removal, and the static-data-provider requirement. Not for everyday assertion writing or PHP < 8.1 projects (stay on phpunit-9-modern). Output: a 9 → 10 migration plan. Pair with phpunit-9-modern for the API conventions that carry forward unchanged.
pytest-async
Async pytest + pytest-asyncio testing: asyncio_mode=auto, SQLite fixtures, httpx.AsyncClient + ASGITransport, unit/integration split, coverage floor, live markers. Use when writing or reviewing async FastAPI/SQLAlchemy tests, or enforcing TDD. Not for production code. Output: offline-green async tests.
python-pro
Modern Python (3.10+) review checklist: typing, dataclasses/pydantic, async-await discipline, logging over print, exception chain, ruff+pyright. Use when writing or reviewing backend Python (async FastAPI/SQLAlchemy). Not for non-backend or pre-3.10 code. Output: typed, ruff/pyright-clean code.
react-18-notes
React 18 (March 2022) signature features and the migration traps from 17 -> 18. Use when writing or reviewing code in a React 18 project, or a package whose package.json constraint is ^18 (or ^18.2). Covers createRoot/hydrateRoot, automatic batching, the opt-in concurrent features (startTransition/useTransition/useDeferredValue), streaming SSR (renderToPipeableStream), the new hooks (useId/useSyncExternalStore/useInsertionEffect), and StrictMode effect double-invocation. Not for application business logic — load when working on React root setup, concurrent-rendering code, or planning a 17 -> 18 (or 18 -> 19) upgrade. Pairs cross-stack with the nextjs modules; React 18.2+ is the floor Next.js 16 supports. Output: version-specific guidance, migration traps, and verification gates.
react-19-notes
React 19 (December 2024) signature features and the breaking-change traps from 18 -> 19. Use when writing or reviewing code in a React 19 project, or planning an upgrade from React 18. Covers Actions and async transitions, the new hooks (useActionState/useOptimistic/useFormStatus and the use() API), ref as a regular prop (forwardRef deprecated), <Context> as a provider, document metadata hoisting, resource preloading (preload/preinit), stable Server Components/Actions, and the removal of long-deprecated APIs (ReactDOM.render/hydrate, propTypes/defaultProps on function components, legacy Context, string refs). Not for application business logic — load when working on React-language code or planning an 18 -> 19 upgrade. Note: React 19 is recommended but NOT required for Next.js 16 (React 18.2+ also works). Output: version-specific guidance, migration traps, and verification gates.
swift-test-strategy
iOS test strategy across Swift Testing (@Test / #expect / #require / parameterized @Test(arguments:) / traits / suites) and XCTest (UI tests, performance metrics), XCUITest automation (XCUIApplication, launch-argument test seams), and swift-snapshot-testing. Defines a 3-layer taxonomy (unit / integration / UI) mapped to iOS test targets, async patterns for actor-isolated services, and RED-first guidance. Use when writing or reviewing iOS tests, choosing Swift Testing vs XCTest, structuring test targets, testing async/actor code, or planning coverage for a feature/bugfix. Not for non-test app code (swift / swiftui / ios-platform); output is RED-first test files plus a 3-layer coverage plan. Requires the swift module.
xcode-build-tooling
Xcode / SwiftLint / xcodebuild / Swift Package Manager tooling — SwiftLint rule-tier strategy and per-package .swiftlint.yml, xcodebuild -scheme/-destination build & test vs swift build / swift test for SPM packages, and scheme / code-signing notes for a no-iCloud capability set. Use when configuring SwiftLint, deciding how to build/test an iOS app or SPM package from the CLI, interpreting the module's post-edit SwiftLint hook or pre-commit build gate, or setting up CI build commands. Not for Swift language questions (swift) or test strategy (swift-test-strategy); output is build/lint commands plus hook and CI config. Requires the swift module; wires post-edit-swiftlint.sh and pre-commit-swift-build.sh.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.