Angular frontend conventions, patterns, and test infrastructure. Use when working with Angular components, services, modules, RxJS, Angular CLI, ng commands, Jasmine tests, Component Harnesses, HttpTestingController, Angular routing, Angular forms, Angular pipes, or Angular dependency injection.
Add a reference-image test to Lodestone, or deliberately update reference images after an intended rendering change. Use when a rendering feature needs image coverage, or a reference-image test fails because rendering was meant to change.
This skill should be used when the user asks about "test doubles", "mocking", "stubbing", "spies", "verifying doubles", "partial doubles", "allow", "receive", "have_received", or needs guidance on isolating tests and mocking dependencies in RSpec.
End-to-end procedure for porting any Flutter/Dart file from the poc_vedaxl app into this React Native monorepo (models, repositories, services, state, widgets, screens, navigation, l10n). Use it whenever you start, continue or review the port of a `lib/**/*.dart` file, when you're asked "port X", "convert this Dart file", "where does <dart file> go", or when you need the Flutter→RN concept mapping (widgets, ChangeNotifier, FutureBuilder, Navigator, ThemeData, sheets, animations, CustomPainter, ARB). For a single screen use flutter-screen-to-rn; for models/repositories use dart-model-to-ts; for tests use flutter-test-to-jest.
Review any engineering task, bug, PRD, or PM proposal when an engineer wants to pressure-test its rationale and choose the next move. Say when its scoped next step is good enough; otherwise recommend a better test or a material decision to resolve. Do not invoke for routine implementation requests.
Automate terminal UI (TUI) apps with agent-tui for testing, inspection, demos, and scripted interactions. Use when automating CLI/TUI flows, regression testing terminal apps, verifying interactive behavior, or extracting structured text from terminal UIs. Also use when asked what agent-tui is, how it works, or to demo it. Do not use for web browsers, GUI apps, or non-terminal interfaces.
Use when a task is well-defined, has an automatic way to verify success (tests, linter, build), and is tedious or large enough to benefit from unattended iteration - e.g. "get all tests passing", "implement this spec end to end", "fix every lint error". Not for one-shot changes or work that needs human judgment mid-stream.
Use when a new font licensing and fallback readiness request needs a bounded work scope to produce a scoped intake card for the font dependency and rights register. Success means owner, objective, permitted sources, acceptance condition, exclusions, and deadline are explicit; every distributed font has a documented use basis and fallback rendering is tested. Use configured search, fetch, read, browser, test, and write capabilities only when relevant and authorized. Record evidence, limit refinement to three focused passes, and seek approval before external or irreversible actions.
Expert VBA code reviewer for Excel/Word/Outlook/PowerPoint macros and add-ins. Use this skill whenever the user pastes VBA and asks for a review, audit, feedback, or help improving it. Also trigger when the user reports a bug, a slow macro, a wrong UDF result, a workbook that won't close, or unexpected Office behavior and shares VBA as context — even without the word "review". Trigger on phrases like "what's wrong with this macro", "can you check this VBA", "is this correct", "why is this slow", "review my macro", or any paste of VBA followed by a question. Prioritize correctness over style: check Option Explicit, declaration bugs, error handling, range/array performance, and application-state hygiene before anything cosmetic. Because VBA has no external CI, a review must also say how to *verify* a fix inside the host Office app. For C#/VB.NET Office add-ins use VSTO-review instead; to author the tests a review asks for, use vba-code-test-writing.
Use when working in the Vivreal Portal (Vivreal_Portal_Mobile), a Next.js 16 App Router web app despite the "Mobile" name. Covers the proxy-route layer (createProxyHandler factory vs hand-written), the three-tier API client rule (createAuthAxios / publicAxios / fetch), the four backend upstreams, auth cookies (token + active_ctx), the retired in-app agent (AI is MCP only), feature flags, PWA shell traps, offline honesty, the Socials and Channels screens, SSR-safe patterns, and e2e/unit test gates. Triggers on: portal, Vivreal_Portal_Mobile, proxy route, createAuthAxios, createProxyHandler, active_ctx, AuthContext, edge runtime, basePath /app, site analytics dashboard, social hub, AI FAB, agent drawer, EditPlan, feature flags, Instagram inbox, reach-out console (formerly cold-call), demo-account claim flow, site template picker, Studio editor, dashboard insights, account deletion, e2e testing, coverage map. Source of truth: Vivreal_Portal_Mobile CLAUDE.md at origin/stable (recount src/app/api/proxy/).
Repeatable procedure for validating and honestly documenting a Bukkit/Paper/Folia plugin's compatibility when a new Minecraft/Paper version ships — research-first API checking, build strategy, runtime verification, and maintaining a tested-vs-assumed compatibility matrix. Use whenever a new Minecraft or Paper version is released, the user asks "does my plugin support version X", a paper-api dependency bump is considered, or release notes need a compatibility claim.
Player-facing input signs for browser games: WASD, vehicles, flight, FPS mouse-look, and the #1 failure mode (inverted A/D). Mandatory control self-tests and a tiny test interface so you can verify A turns left before shipping. Load for ANY game with movement, steering, flying, driving, tanks, boats, mechs, drones, planes, karts, third-person follow cams — not only racing. Triggers on "controls", "WASD", "inverted", "steer", "flight", "airplane", "kart", "vehicle", "yaw", "roll", "pitch", "mouse look", "A/D".
User describes a research question — what they're trying to predict or explain, what variables are involved, and what the study design looks like — and wants a recommended statistical model with a hierarchical test sequence from null to fully specified. Use this skill whenever the user asks which statistical test or model to use, wants to plan an analysis, needs help choosing between ANOVA/regression/mixed effects/CCA/factor analysis, or describes outcome and predictor variables and wants a structured modelling approach.
基于视觉的浏览器自动化，使用 Midscene 驱动无头 Puppeteer 浏览器。
纯截图驱动，无需 DOM 或无障碍标签，可操控页面上任何可见元素。
运行在无头浏览器中，不会接管用户的鼠标键盘，用户可同时使用电脑。
适用于：浏览网页、抓取数据、填写表单、点击按钮、测试前端 UI、截图网页、自动化 Web 工作流。
触发词：浏览器自动化, 打开网页, 网页截图, 抓取数据, 填写表单, 测试网页, 前端测试,
browser automation, open URL, web scraping, fill form, test web app, take screenshot,
browse website, check page, QA frontend, end-to-end test
Simulate 5 buyer personas reading any landing page, email, ad, sales page, proposal, or offer, section by section, then report who would buy, who would bounce, where friction clusters, and what to fix first. Generates a visual HTML report with conversion score, buyer-by-buyer walkthroughs, section-by-section friction map, prioritized fixes, and rewrite suggestions. MANDATORY TRIGGERS: 'buy or bounce this', 'run buy or bounce', 'simulate buyers', 'buyer sim this', 'would they buy this', 'who would bounce', 'test this before launch'. STRONG TRIGGERS: 'review this like a buyer', 'where would people drop off', 'why is this not converting', 'pressure test this landing page', 'simulate conversion objections'. Do NOT trigger on requests for brand voice editing, general proofreading, or analytics interpretation after a live launch.
Builds @haustle/notion-orm from source, links it into orm-testing, and runs notion sync to refresh generated Notion artifacts. Use when the user wants to ship ORM changes into the local orm-testing sandbox, says build and sync, link and notion sync, or mentions orm-build-sync.