convez.ai
AgenteAutomaçãoAvançado
Deepankar1993

openAssistant

Deepankar1993/openassistant

Assistente de IA pessoal cross-platform com memória, skills e multi-agent

Ver no GitHub
InicialPluginScore · de 0 a 100

Assistente de IA pessoal unificado em Rust que combina OpenHumans, Hermes Agent e OpenClaw. Oferece agent autoaprendizável com memória persistente, gateway multi-plataforma (Discord, Telegram, Slack, Signal), ferramentas integradas (navegação, shell, visão, cron) e lona visual em tempo real.

Instalação

$npx convez-ai Deepankar1993/openassistant

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
0
Acompanhando
0
Saúde
28%
Último commit
há 4 meses
Versão
v0.3.0
Issues abertas
0
Contribuidores
1

Casos de uso

  • Automação de tarefas repetitivas com cron
  • Assistência conversacional contínua em múltiplos canais
  • Gerenciamento de memória persistente entre sessões
  • Exploração de dados pessoais via OpenHumans

Veredito

Projeto ambicioso que ambiciona ser uma plataforma completa de agente pessoal. Apesar da proposta interessante, carece de documentação clara sobre status de manutenção, estabilidade em produção e falta exemplos práticos de use-cases reais. Recomendado para experimentos e prototipagem.

README

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

🦞 openAssistant

Your own personal AI assistant. Any OS. Any Platform.

A unified personal AI assistant in Rust, combining the best of:

  • OpenHumans — Data exploration, sharing, projects, community
  • Hermes Agent — Self-improving agent with memory, skills, cron, multi-platform messaging
  • OpenClaw — Multi-channel inbox, voice wake, live canvas, DM pairing, companion apps

Features

🧠 Self-Improving Agent

  • Memory: SQLite-backed persistent memory with FTS5 search
  • Skills: Auto-create skills from experience, improve during use
  • Session Search: Full-text search across all past conversations
  • User Modeling: Builds a deepening model of who you are

💬 Multi-Platform Gateway

  • Discord, Telegram, Slack, Signal, WebChat
  • DM pairing mode for security (OpenClaw-style)
  • Cross-platform conversation continuity
  • Voice support (planned)

🛠 Tools

  • Browser: Web search and site browsing
  • Shell: Execute commands on the host
  • File: Read, write, edit files
  • Vision: Analyze images via Gemini CLI directly
  • Memory: Store and search persistent memories
  • Cron: Schedule automated tasks

📊 OpenHumans Integration

  • Data source management
  • Project creation and collaboration
  • Member management
  • Activity feeds

🎨 Live Canvas

  • Agent-driven visual workspace
  • Render text, charts, and HTML
  • Real-time updates

🔒 Security

  • DM pairing (unknown senders get a code)
  • Allowlist-based access control
  • Secure defaults

✨ Onboarding

  • Interactive wizard (openassistant onboard)
  • Guided setup of providers, channels, and skills
  • Works on macOS, Linux, and Windows

Quick Start

# Install
cargo install open-assistant

# Run onboarding wizard
openassistant onboard

# Start chatting
openassistant chat

# Start the gateway
openassistant gateway

# Check status
openassistant status

# Run diagnostics
openassistant doctor

Commands

CommandDescription
chatInteractive CLI chat
gatewayStart messaging gateway
onboardRun setup wizard
configView/edit configuration
statusShow status
doctorRun diagnostics
updateUpdate to latest version

Configuration

Configuration is stored in ~/.openassistant/config.yaml:

general:
  data_dir: ~/.openassistant
  log_level: info

model:
  provider: openrouter
  model: openrouter/owl-alpha
  api_key: your-api-key
  api_base: https://openrouter.ai/api/v1

gateway:
  discord_token: ""
  telegram_token: ""
  slack_token: ""
  webhook_port: 3000
  dm_policy: pairing

memory:
  db_path: ~/.openassistant/memory.db
  fts_enabled: true

skills:
  dirs:
    - ~/.openassistant/skills
  auto_create: true

security:
  dm_pairing: true
  allow_from: []

vision:
  provider: gemini-cli
  gemini_path: gemini

Architecture

src/
├── main.rs           # CLI entry point
├── core/             # Agent engine, sessions, context
├── gateway/          # Multi-platform messaging
├── memory/           # Persistent memory (SQLite + FTS5)
├── skills/           # Skills system with auto-creation
├── cron/             # Scheduled automations
├── tools/            # Browser, shell, file, vision
├── platforms/        # OpenHumans data sources
├── canvas/           # Live canvas renderer
├── security/         # DM pairing, allowlist
├── onboarding/       # Setup wizard
├── ui/               # TUI chat interface
└── config/           # Configuration management

Vision Tool

openAssistant uses Gemini CLI directly for vision analysis — no API key needed, just install Gemini CLI:

# Install Gemini CLI
npm install -g @anthropic/gemini-cli

# Or on Windows
iex (irm https://raw.githubusercontent.com/google/gemini-cli/main/scripts/install.ps1)

# Then the vision tool works automatically
open assistant → "Analyze this image: /path/to/image.jpg"

License

MIT

Ver o original no GitHub

Bom para

Relacionados