manhthien2005

Verify WCAG 2.1 AA compliance by analyzing HTML structure, ARIA usage, contrast, keyboard patterns, and semantic markup. Runs before completion in FE Designer mode.

Guide the design of beautiful, accessible, systematic UI components. Auto-triggers when creating or restyling components in FE Designer mode.

Audit design system consistency — check tokens, spacing, colors, typography usage across the codebase. Triggers when modifying shared styles or on user request.

Self-critique visual output before completion. Evaluates visual hierarchy, CTA clarity, cognitive load, spacing consistency, and layout priority. MUST run before attempt_completion in FE Designer mode.

MANDATORY: Run validation tools after EVERY code change. Do not finish until code is error-free.

Auto-enforce coding standards after every code change in Code mode. Checks file size, nesting, naming, error handling, and anti-patterns.

Monitor and manage context window usage to prevent overflow and optimize token efficiency.

Capture lessons learned, patterns, and user preferences after complex sessions to improve agent behavior over time.

Use when writing or modifying Express + Prisma backend code in VSmartwatch admin (HealthGuard/backend). Reference patterns for Prisma client singleton, JWT auth, route/controller/service layering, error handling, Postgres patterns, Socket.IO realtime, AWS S3, and Jest testing.

Use when writing or modifying FastAPI backend code in VSmartwatch (health_system/backend, healthguard-model-api, Iot_Simulator_clean/api_server). Reference patterns for router/service/repository layering, Pydantic v2, async DB, internal-secret auth, error handling, sanitized responses, and pytest.

Use when writing or modifying Flutter/Dart code in VSmartwatch mobile app (health_system/lib). Reference patterns for Riverpod state management, dio HTTP, repository pattern, GoRouter navigation, FCM (push only), elderly UX, and widget testing.

Use when about to claim work is complete, fixed, or passing. Requires running verification commands and reading actual output BEFORE making any success claims. Evidence before assertions.

Ultra-compressed Vietnamese mode. ONLY activate when the user types "/caveman" or says "bật caveman", "caveman mode", "ngắn gọn cực", "tiết kiệm token". Off by default. Output stays Vietnamese; only the response style changes.

Five-axis code review framework. Use when reviewing your own changes before merge, or reviewing PR. Evaluates correctness, readability, architecture, security, performance.

Behavioral guidelines to reduce common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable success criteria. Applies to ALL code work in VSmartwatch.

Manage screen spec lifecycle for VSmartwatch mobile app — generate from UC/SRS, edit, cross-link, and maintain the Screen Index. Use when user says "list all screens", "quản lý màn hình", "scan screens", "generate screen spec", "kiểm tra screen index", "sync screens", or "TASK <subcommand>".

MANDATORY: Run appropriate validation tools after EVERY code change. Do not finish a task until the code is error-free.

Implements Manus-style file-based planning to organize and track progress on complex tasks. Creates task_plan.md, findings.md, and progress.md. Use when asked to plan out, break down, or organize a multi-step project, research task, or any work requiring 5+ tool calls. Supports automatic session recovery after /clear.

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

Legacy compatibility alias for task-router-lite. Keeps older codex-max references working while using a thin Phase PLAN routing model.

Thin Phase PLAN router for codex-max. Select at most one primary execution skill and one optional secondary skill from the installed local skill surface.

Use when a significant technical decision should be recorded with context, alternatives, and consequences so the reasoning survives beyond the current session.

Use before creative or constructive work when requirements or solution shape are still unclear. Turns vague ideas into confirmed designs before implementation begins.

Optional browser automation workflow for interactive browser tasks, Playwright-style automation, and stable scraping/testing patterns.

Write and organize unit tests for functions, methods, and classes using `package:test`. Use when creating new logic or fixing bugs to ensure code remains correct and regression-free.

Uses get_runtime_errors and lsp to fetch an active stack trace, locate the failing line, apply a fix, and verify resolution via hot_reload.

Define and generate mock objects for external dependencies using `package:mockito` and `build_runner`. Use when unit testing classes that depend on complex external services like APIs or databases.

Execute `dart analyze` to identify warnings and errors, and use `dart fix --apply` to automatically resolve mechanical lint issues. Use during development to ensure code quality and before committing changes.

Configures Flutter Driver for app interaction and converts MCP actions into permanent integration tests. Use when adding integration testing to a project, exploring UI components via MCP, or automating user flows with the integration_test package.

Implement a component-level test using `WidgetTester` to verify UI rendering and user interactions (tapping, scrolling, entering text). Use when validating that a specific widget displays correct data and responds to events as expected.

Architects a Flutter application using the recommended layered approach (UI, Logic, Data). Use when structuring a new project or refactoring for scalability.

Fixes Flutter layout errors (overflows, unbounded constraints) using Dart and Flutter MCP tools. Use when addressing "RenderFlex overflowed", "Vertical viewport was given unbounded height", or similar layout issues.

Audit and review current JIRA sprint backlog progress by scanning _SPRINT.md and STORIES.md files. Generates progress tracking reports with completion rates and risk analysis. Triggers: check progress, review backlog, tiến độ sprint, kiểm tra backlog, sprint review, audit tasks.

Quét mã nguồn thực tế và cập nhật Project_Structure.md + summaries/ để PM_REVIEW luôn khớp với code hiện tại. Dành cho PM/AI agent quản lý cấu trúc dự án HealthGuard. Kích hoạt khi: "check project", "cập nhật cấu trúc", "kiểm tra source", "project thay đổi chưa?", "sync lại PM", "update summaries", "review structure", "kiểm tra project", "project còn đúng không?", "check source", "update structure".

Perform a detailed evaluation of a specific feature — analyze actual source code against architecture standards, verify SRS compliance, assess implementation quality, and cross-check JIRA Stories for the project development team. Use when user says "review code", "đánh giá chi tiết", "evaluate feature", "check implementation", "review module", "review chức năng", "kiểm tra code", "detailed review", "danh gia chi tiet", "xem code giúp", "chấm điểm code", "review feature", "score this module". DO NOT use for: PR reviews, debugging, writing new code, generating test cases.

Implement proven backend architecture patterns including Clean Architecture, Hexagonal Architecture, and Domain-Driven Design. Use when architecting complex backend systems or refactoring existing ...

Automatically generate or draft standards-compliant SRS (Software Requirements Specification) and SDD (Software Design Document - IEEE-1016-2009) from source code and requirements. Utilizes a multi-agent brainstorming model with Architect and BA roles to cross-review and produce the documentation. Use when the user requests "write SRS", "write SDD", "create IEEE-1016-2009 template", "generate design document", "generate system design document".

Comprehensive toolkit for product managers including RICE prioritization, customer interview analysis, PRD templates, discovery frameworks, and go-to-market strategies. Use for feature prioritizati...

Specialized agent for designing, building, reviewing, and managing Flutter mobile app UI. Use when user says "design screen", "thiết kế màn hình", "code UI", "build UI", "review screen", "kiểm tra giao diện", "list all screens", "quản lý màn hình", "plan for module", "SOS screen design", "dashboard layout", or shorthand like "plan login page", "build home UI". Supports 4 modes: PLAN, BUILD, REVIEW, TASK.

Act as an elite Project Manager to analyze UC, SRS, and system overview then generate structured Sprint Plans, Epics, and Stories following the established JIRA template structure. Triggers: sắp xếp task, dự toán công việc, tạo task list, lên plan, lập kế hoạch, sprint planning, create sprint, estimate workload, backlog grooming, epic breakdown, story mapping, PM review planning, chia task, phân công việc.

Safely modify the Drizzle schema and create a migration for FixNow Postgres. Validates the change against docs/database-schema.md, runs drizzle-kit generate then migrate, and updates the schema doc. Use when user says "thêm field X vào users", "đổi enum status", "add Review table", "tạo migration".

Bootstrap the FixNow Next.js project from scratch — install Next.js, Tailwind, Drizzle ORM, NextAuth, shadcn/ui, and set up the base folder structure. Use this skill ONCE when starting a fresh project, or to verify all base dependencies are present. Triggers when user says "khởi tạo dự án", "init FixNow", "scaffold project", "setup base".

Run a responsive QA pass on a FixNow page or component across desktop, tablet, and mobile Chrome viewports. Verifies no horizontal scroll, touch targets, Vietnamese diacritics, layout shifts, and Lighthouse mobile score. Use when user says "kiem tra responsive", "test trang X", "check tren mobile/tablet", "qa giao dien".

Scaffold a Next.js App Router API route (route.ts) for FixNow with zod validation, auth check via NextAuth, Drizzle ORM access, and proper error handling. Use when user says "tạo API endpoint X", "scaffold POST /api/appointments", "thêm route handler GET /api/parts".

Scaffold a new Next.js App Router page following FixNow conventions — correct route group, Server vs Client component split, layout integration, mobile-first Tailwind. Use when user says "tạo trang X", "scaffold page /pricing", "thêm route /track".

Manage db/seed.ts for FixNow with realistic Vietnamese fixture data for services, parts, admin user, and sample appointments using Drizzle. Use when user says "seed data", "tạo dữ liệu mẫu", "reset demo data", "thêm linh kiện vào DB".

Add or update shadcn/ui components for FixNow by running the official shadcn CLI. Never copy component source from the shadcn-ui/ui repo by hand. Use this when the user says "them component button", "cai shadcn select", "shadcn table", "init shadcn", or whenever a required component is missing from components/ui/.

UI/UX design intelligence for web and mobile. Includes 50+ styles, 161 color palettes, 57 font pairings, 161 product types, 99 UX guidelines, and 25 chart types across 10 stacks (React, Next.js, Vue, Svelte, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui, and HTML/CSS). Actions: plan, build, create, design, implement, review, fix, improve, optimize, enhance, refactor, and check UI/UX code. Projects: website, landing page, dashboard, admin panel, e-commerce, SaaS, portfolio, blog, and mobile app. Elements: button, modal, navbar, sidebar, card, table, form, and chart. Styles: glassmorphism, claymorphism, minimalism, brutalism, neumorphism, bento grid, dark mode, responsive, skeuomorphism, and flat design. Topics: color systems, accessibility, animation, layout, typography, font pairing, spacing, interaction states, shadow, and gradient. Integrations: shadcn/ui MCP for component search and examples.

Use before claiming any BetterMe task is done, or when verifying a change — the exact 4-gate sequence (typecheck, lint, vitest, build), Windows/PowerShell/pnpm specifics, single-test and dev-bypass usage, and what counts as evidence.

Use when writing or editing any user-facing Vietnamese copy — pet/companion voice lines, toasts, greetings, milestone text, or friend/garden messaging. Covers the no-guilt and no-comparison invariants, host vs guest voice, per-species tone, and the test that enforces them.

Use when editing supabase/schema.sql, RLS policies, RPCs, or the server actions in src/lib/server/** — covers owner-only RLS, SECURITY DEFINER cross-user reads, RPC-only writes, idempotent schema patterns, the append-only food ledger, and the never-throw server-action contract.

Use when working in src/lib/sync/**, on offline/local-first behavior, multi-device merge, conflict resolution, localStorage state, or upload/import — covers the merge laws (monotonic max, per-cell LWW, append-only ledgers, tombstones), seed-fiction provenance, and the derived-cache rule.

Create beautiful, accessible user interfaces with shadcn/ui components (built on Radix UI + Tailwind), Tailwind CSS utility-first styling, and canvas-based visual designs. Use when building user interfaces, implementing design systems, creating responsive layouts, adding accessible components (dialogs, dropdowns, forms, tables), customizing themes and colors, implementing dark mode, generating visual designs and posters, or establishing consistent styling patterns across applications.

Comprehensive design guide for web, mobile, and desktop applications. Contains 67 styles, 161 color palettes, 57 font pairings, 99 UX guidelines, and 25 chart types across 22 technology stacks.

Use when designing, building, or redesigning frontend landing pages, portfolios, editorial sites, and marketing experiences that need a distinctive, production-ready visual direction instead of templated AI aesthetics.

Use when designing, building, reviewing, or improving UI/UX for web, mobile, or desktop interfaces, especially when choosing styles, palettes, typography, charts, layout, accessibility, motion, landing patterns, or stack-specific implementation guidance.

Quality-first orchestration for Codex coding workers: complete specs, contract-safe decomposition, adversarial self-review, ground-truth reports, and evidence-based QC. Use before delegating implementation or reviewing worker output.