convez.ai
PluginDados & AnalyticsAvançado
jbouder

Starkiller - BI Generativo com IA

jbouder/starkiller

Gere dashboards e visualizações com IA a partir de perguntas naturais

Ver no GitHub
InicialPluginScore · de 0 a 100

Ferramenta experimental de Business Intelligence que usa LLMs para gerar visualizações de dados em tempo real a partir de consultas em linguagem natural. Em vez de dashboards pré-construídos, você faz perguntas sobre seus dados e recebe gráficos e insights visuais gerados dinamicamente pela IA.

Instalação

$npx convez-ai jbouder/starkiller

1 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

Estrelas
0
Forks
1
Acompanhando
0
Saúde
42%
Último commit
há 9 meses
Versão
—
Issues abertas
6
Contribuidores
2

Casos de uso

  • Análise exploratória de dados sem pré-definir dashboards
  • Geração dinâmica de gráficos e relatórios visuais sob demanda
  • Consultas conversacionais sobre métricas e KPIs de negócio
  • Prototipagem rápida de visualizações de dados

Veredito

Projeto ambicioso que reimagina BI tradicional com IA gerativa, integrando Claude/Gemini com FastAPI e React. Experimento promissor para análise conversacional, mas ainda em fase conceitual — requer setup local ou Docker e chaves de API para operacional.

README

A tradução para o português deste README ainda está na fila.

Starkiller

An experimental Generative BI (Business Intelligence) tool that leverages Large Language Models to generate data visualizations on the fly.

Overview

Starkiller reimagines traditional BI dashboarding by using AI to dynamically create data visualizations based on natural language queries. Instead of pre-built dashboards, users can ask questions about their data and receive intelligent, contextual visualizations generated in real-time.

Features

  • AI-Powered Visualization: Generate charts, graphs, and dashboards using natural language
  • Dynamic Analysis: Ask questions and get instant visual insights
  • Flexible Data Sources: Connect to various data sources and databases
  • Interactive Dashboards: Explore and refine visualizations interactively
  • Smart Recommendations: AI suggests relevant visualizations based on your data

Tech Stack

FrontendBackend
React 19 + TypeScriptPython FastAPI
ViteSQLAlchemy (async)
Tailwind CSSAnthropic Claude & Gemini
shadcn/uiPandas/NumPy
RechartsAlembic + PostgreSQL

Quick Start (Docker Compose)

The easiest way to run Starkiller is with Docker Compose, which starts the UI, API, and PostgreSQL database together.

Prerequisites

1. Clone and Configure

git clone https://github.com/yourusername/starkiller.git
cd starkiller

# Configure the API environment
cp api/.env.example api/.env

Edit api/.env and configure your preferred LLM provider:

# Supported providers: anthropic, gemini
LLM_PROVIDER=anthropic

# If using Anthropic
ANTHROPIC_API_KEY=sk-ant-your-key-here

# If using Gemini
GEMINI_API_KEY=your-gemini-key-here

2. Start the Stack

docker compose up --build

3. Initialize the Database

In a new terminal, run migrations and seed sample data:

# Run database migrations
docker compose exec api alembic upgrade head

# Seed external sample databases
docker compose exec api python scripts/seed_connection_data.py

# Seed application data sources
docker compose exec api python scripts/seed_data_sources.py

# Seed sample dashboards
docker compose exec api python scripts/seed_dashboards.py

4. Access the Application

ServiceURL
UIhttp://localhost:3000
APIhttp://localhost:8000
API Docs (Swagger)http://localhost:8000/docs
API Docs (ReDoc)http://localhost:8000/redoc

Local Development (Without Docker)

For development without Docker, you'll need to run the frontend and backend separately.

Prerequisites

Backend Setup

cd api

# Create and activate virtual environment
python -m venv .venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate

# Install dependencies
pip install -e ".[dev]"

# Configure environment
cp .env.example .env
# Edit .env and configure your LLM provider
# Example for Gemini:
# LLM_PROVIDER=gemini
# GEMINI_API_KEY=your-key-here

# Run migrations
alembic upgrade head

# Start the API server
uvicorn main:app --reload

Frontend Setup

cd ui

# Install dependencies
npm install

# Start the development server
npm run dev

Seed Sample Data (Local)

With PostgreSQL running locally:

cd api
source .venv/bin/activate

# Set admin connection for creating external databases
export ADMIN_DB_URL=postgresql://postgres:postgres@localhost:5432/postgres

# Seed in order
python scripts/seed_connection_data.py  # External sample databases
python scripts/seed_data_sources.py     # Application data sources
python scripts/seed_dashboards.py       # Sample dashboards

Local URLs

ServiceURL
UIhttp://localhost:5173
APIhttp://localhost:8000

Common Commands

Running Tests

cd api
pytest                           # Run all tests
pytest --cov=. --cov-report=html # With coverage report
pytest tests/unit/               # Unit tests only
pytest -k "test_name"            # Run specific test

Building for Production

cd ui
npm run build    # Build frontend
npm run preview  # Preview production build

Adding UI Components

cd ui
npx shadcn@latest add <component-name>

Linting

cd ui
npm run lint

Environment Variables

API (api/.env)

VariableDescriptionDefault
LLM_PROVIDERLLM provider (anthropic, gemini)anthropic
ANTHROPIC_API_KEYAnthropic API key(required if using anthropic)
ANTHROPIC_MODELClaude model to useclaude-sonnet-4-5-20250929
GEMINI_API_KEYGoogle Gemini API key(required if using gemini)
GEMINI_MODELGemini model to usegemini-2.0-flash
DATABASE_URLDatabase connection stringpostgresql+asyncpg://postgres:postgres@localhost:5432/starkiller
ENVIRONMENTdevelopment, staging, productiondevelopment
DEBUGEnable debug modetrue
HOSTAPI host0.0.0.0
PORTAPI port8000
CORS_ORIGINSAllowed CORS origins["http://localhost:5173"]

License

This project is licensed under the GNU General Public License v3.0 - see the LICENSE.md file for details.

Ver o original no GitHub

Bom para

Relacionados