Retirement AI Dashboard
btholath/retirement-ai-dashboard
Dashboard de aposentadoria com IA e MCP para gestão de anuidades
Ver no GitHubDashboard de aposentadoria assistido por IA, construído com Next.js 16, TypeScript e Claude Code. Demonstra workflows agentic, integração MCP e engenharia de prompts para gestão de anuidades de aposentadoria e contas de participantes.
Instalação
npx convez-ai btholath/retirement-ai-dashboard1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Gerenciar e visualizar contas de participantes de aposentadoria
- Demonstrar arquitetura agentic com Claude Code e MCP
- Exemplo de engenharia de prompts para geração e review de código
Veredito

Projeto robusto que combina Next.js 16, TypeScript strict e integração MCP para criar um dashboard de domínio financeiro. Serve como referência prática de arquitetura agentic e workflows de IA aplicados a desenvolvimento full-stack.
README
A tradução para o português deste README ainda está na fila.
retirement-ai-dashboard
AI-assisted retirement dashboard built with Next.js 16, TypeScript, and Claude Code. Demonstrates agentic workflows, MCP integration, and prompt engineering for retirement annuity and participant account management.
What this demonstrates
| Capability | Implementation |
|---|---|
| Claude Code | End-to-end feature generation, refactoring, and code review via CLI |
| Prompt Engineering | Structured prompts for code generation, review, and architectural guidance |
| Agentic Architecture | CLAUDE.md + AGENTS.md rules drive consistent AI-assisted development |
| Next.js 16 | App Router, React Server Components, file-based routing |
| TypeScript | Strict mode, interfaces, runtime type guards — no unsafe casts |
| Retirement Domain | Participant accounts, annuity balances, projection-ready data model |
Tech stack
- Framework: Next.js 16 (App Router)
- Language: TypeScript 5 (strict)
- Styling: Tailwind CSS v4
- AI tooling: Claude Code (Anthropic) with MCP integration
- Runtime: Node.js 20+
Project structure
├── app/
│ ├── layout.tsx # Root layout with Geist font
│ └── page.tsx # Homepage — server component
├── src/
│ ├── components/
│ │ └── ParticipantCard.tsx # Reusable participant card
│ ├── data/
│ │ └── participants.json # Seed data
│ ├── lib/
│ │ ├── formatters.ts # formatCurrency, pluralize
│ │ └── participants.ts # getParticipants() with type guard
│ └── types/
│ └── participant.ts # Participant interface
├── scripts/
│ └── push-to-github.sh # Idempotent GitHub push script
├── CLAUDE.md # AI architect rules and domain context
└── AGENTS.md # Agentic Next.js conventions
Local setup
# 1. Install dependencies
npm install
# 2. Start the dev server
npm run dev
# 3. Open in browser
open http://localhost:3000
Built with Claude Code
This project was built interactively using Claude Code, Anthropic's AI-powered CLI. The workflow demonstrates:
- Feature generation — homepage, types, services, and components generated from a structured prompt
- Code review — staff-engineer-level review identifying unsafe casts, module-scope optimizations, and naming conventions
- Refactoring — type guard replacing
ascast, formatter extracted tolib/,Intl.NumberFormathoisted to module scope - Git workflow — conventional commits created and pushed via an AI-generated idempotent shell script
See 06_prompt_engineering.md for the exact prompts used.
Roadmap
- Retirement projection calculator (
src/lib/projections.ts) - Annuity processing service
- REST API routes (
app/api/participants/) - Vitest unit tests for all lib functions
- GitHub Actions CI (typecheck + lint)
- Participant detail page with projection chart
Related
- Blog: blog.retiriq.com
- GitHub: @btholath