Skip to content

About

AI-powered personal assistant built with LangGraph, PostgreSQL, Browser MCP, WhatsApp, Gmail, and Calendar integrations, featuring persistent memory, automation, and document generation.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

Agent Vortex V2

Agent Vortex is a fully Dockerized, personal AI operating system built with LangGraph, PostgreSQL, MCP servers, and modern LLM tooling.

It seamlessly integrates multiple disparate tools into a single autonomous workflow capable of reasoning, planning, executing, and persisting context across sessions.

🌟 Key Features

  • Browser Automation: Powered by Playwright MCP. Can navigate to URLs, extract content, capture snapshots, interact with web pages, and manage tabs.
  • WhatsApp Automation: Powered by a built-in WhatsApp MCP sidecar. Allows the agent to send WhatsApp messages natively.
  • Google Workspace Integration: Connects to Gmail and Google Calendar to read/send emails and manage scheduling.
  • Persistent Memory Stack: Short-Term Memory (STM) for conversational context and Long-Term Memory (LTM) stored in PostgreSQL for permanent context injection.
  • File Generation: Capable of generating robust PDF documents and Excel spreadsheets on the fly.
  • Human-In-The-Loop (HITL): Built-in risk assessment and approval workflows before executing dangerous actions.

🏗️ Architecture

START
  ↓
Planner Node (Designs execution steps)
  ↓
Risk Assessment (Evaluates action danger level)
  ↓
Approval (Optional HITL for high-risk plans)
  ↓
Executor (Runs the tools using Browser-First Logic)
  ↓
END

🚀 Quick Start (Dockerized)

Agent Vortex is now fully Dockerized! The repository contains everything you need to run the Agent, the PostgreSQL database, and the WhatsApp MCP sidecar seamlessly.

1. Prerequisites

  • Docker and Docker Compose installed on your host machine.
  • An API key for your chosen LLM (default is via OpenRouter).

2. Clone the Repository

git clone https://github.com/CherukuriPavanKumar/AgentVortex.git
cd AgentVortex

3. Environment Setup

Copy the example environment file:

cp .env.example .env

Edit your .env file:

  1. Provide your OPENROUTER_API_KEY.
  2. Generate a secure token for the WhatsApp MCP sidecar:
    openssl rand -hex 32
    Paste this value into WHATSAPP_MCP_TOKEN.

4. Optional: Google Integrations

If you wish to use the Gmail or Google Calendar features:

  1. Generate your credentials.json from the Google Cloud Console.
  2. Run the local script to authenticate and generate token.pickle.
  3. Uncomment the volume mounts for these files in docker-compose.yml.

5. Launch the System

Run Docker Compose to build and start the ecosystem:

docker compose up -d --build

6. Pair WhatsApp (Required for WhatsApp Tools)

The WhatsApp MCP server runs securely isolated in its container, requiring a one-time QR code scan from your phone.

  1. Ensure the system is running (docker compose ps).
  2. Open your host machine's browser and navigate to: http://localhost:8765/pair
  3. Open WhatsApp on your phone → Settings → Linked Devices → Link a Device.
  4. Scan the QR code shown on the screen.

Security Note: This pairing page is bound strictly to 127.0.0.1 on your host. It is not exposed to your local network.

7. Interact with the Agent

Attach to the agent container's terminal interface:

docker attach agentvortex-agent-1

(Tip: To detach without stopping the container, press Ctrl+p, then Ctrl+q).

Any PDFs or Excel spreadsheets generated by the agent will appear automatically in your host machine's generated_pdfs/ or generated_excels/ directories via Docker volume mounts.


🛠️ Available Tools

Tool Group Tools
Core terminal_tool, generate_pdf, generate_excel
Browser browser_navigate, browser_read, browser_interaction, browser_tabs
WhatsApp whatsapp_send_message
Google gmail_read, gmail_send, calendar_read, calendar_write

📦 Current Status

Current version: v0.2.0

Implemented in V2:

  • ✅ Complete Dockerization (Agent + Postgres + MCPs)
  • ✅ Secure WhatsApp MCP Pairing via Docker Networking
  • ✅ Browser MCP integration (Playwright)
  • ✅ LLM Tooling & Context Injection
  • ✅ Persistent Conversations (PostgreSQL)

Future Roadmap:

  • Agent UI / Frontend Dashboard
  • Improved semantic vector memory architecture
  • Multi-agent delegation workflows

📄 License

MIT License

About

AI-powered personal assistant built with LangGraph, PostgreSQL, Browser MCP, WhatsApp, Gmail, and Calendar integrations, featuring persistent memory, automation, and document generation.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages