Family AI Agent

Multi-agent AI system with memory and autonomous capabilities. Built with TypeScript, LangGraph.js, and OpenRouter.

Features

• Multi-Agent Architecture: Supervisor agent routes tasks to specialized agents

  • 🗣️ Kakak (Chat): General conversation and Q&A
  • 🔍 Researcher: Web search and information gathering
  • 💻 Coder (Adik): Code generation, debugging, and review
  • ⚙️ Automator: Task scheduling and workflow automation

• Memory System:

  • Conversation memory (session context)
  • Long-term memory (vector search with pgvector)
  • Knowledge base (RAG for documents)

• Safety Features:

  • Input validation and content filtering
  • Output sanitization and PII detection
  • Audit logging
  • Sandboxed code execution (Docker)

Quick Start

Prerequisites

• Node.js 20+
• Docker & Docker Compose
• OpenRouter API key

Installation

# Clone and install dependencies
npm install

# Copy environment file and configure
cp .env.example .env
# Edit .env with your OPENROUTER_API_KEY

# Start database
docker-compose up -d

# Run the CLI
npm run cli chat

CLI Commands

# Interactive chat mode
npm run cli chat

# Single question
npm run cli ask "How do I create a REST API in Node.js?"

# Upload document to knowledge base
npm run cli upload ./document.pdf

# Search knowledge base
npm run cli search "authentication patterns"

# Check system status
npm run cli status

Programmatic Usage

import { quickStart } from 'family-ai-agent';

const ai = await quickStart();

// Ask questions
const response = await ai.ask("Write a function to calculate fibonacci");
console.log(response);

// Store memories
await ai.remember("User prefers TypeScript over JavaScript");

// Search memories
const memories = await ai.search("programming preferences");

// Cleanup
await ai.shutdown();

Architecture

┌─────────────────────────────────────────────────────┐
│                  CLI / API Interface                 │
└─────────────────────────────────────────────────────┘
                          │
┌─────────────────────────────────────────────────────┐
│              Supervisor Agent (Coordinator)          │
│    ┌──────────┬──────────┬──────────┬──────────┐    │
│    │ Research │  Coding  │Automation│   Chat   │    │
│    │  Agent   │  Agent   │  Agent   │  Agent   │    │
│    └──────────┴──────────┴──────────┴──────────┘    │
└─────────────────────────────────────────────────────┘
                          │
┌─────────────────────────────────────────────────────┐
│                   Memory Layer                       │
│  ┌────────────┬─────────────┬──────────────────┐    │
│  │Conversation│  Long-term  │  Knowledge Base  │    │
│  │  Memory    │   Memory    │     (RAG)        │    │
│  └────────────┴─────────────┴──────────────────┘    │
└─────────────────────────────────────────────────────┘

Configuration

Key environment variables:

| Variable | Description | Default | |----------|-------------|---------| | OPENROUTER_API_KEY | OpenRouter API key | Required | | DEFAULT_MODEL | Default LLM model | anthropic/claude-3.5-sonnet | | DB_HOST | PostgreSQL host | localhost | | ENABLE_CONTENT_FILTER | Enable safety filters | true | | ENABLE_AUDIT_LOGGING | Enable audit logs | true |

Development

# Development mode with hot reload
npm run dev

# Build
npm run build

# Run tests
npm test

# Lint
npm run lint

Project Structure

src/
├── core/
│   ├── agents/          # Agent implementations
│   └── orchestrator/    # LangGraph workflow
├── memory/
│   ├── conversation/    # Session memory
│   ├── longterm/        # Vector memory
│   └── knowledge-base/  # RAG system
├── safety/
│   └── guardrails/      # Input/output filtering
├── llm/                 # OpenRouter client
├── api/                 # REST API (future)
└── cli/                 # CLI interface

License

MIT