🤖 Smart Workflow Automation

Transform plain English into powerful automations with AI + MCP + Google Integration

⚠️ Copyright Notice This project is developed and maintained by @jayyycodes. All rights reserved. See LICENSE for usage terms. Unauthorized redistribution or commercial use without permission is prohibited.

✨ What Is This?

Smart Workflow Automation lets you create complex automations using plain English. Powered by a multi-LLM AI engine with dynamic code generation, a full Model Context Protocol (MCP) server, and Google OAuth integrations, it understands your intent, builds workflows automatically, and can even generate custom code on-the-fly for capabilities it doesn't have built-in.

Just say:

"Fetch Bitcoin price and email me via Gmail" → ✅ Done
"Check AAPL stock every 5 minutes and save to Google Sheets" → ✅ Done
"Email me top HackerNews stories every morning" → ✅ Done
"Fetch top stories from Times of India and email me a summary" → ✅ Dynamic!
"Scrape trending GitHub repos and email me" → ✅ Dynamic!
"Create a calendar event for team standup tomorrow at 10am" → ✅ Done
"Fetch latest NDTV headlines and email me" → ✅ Dynamic!

No coding required. Just natural language. If a built-in tool doesn't exist, the AI writes one automatically.

🎯 Live Demo

Try it yourself: Live App

🚀 Features

🤖 Multi-LLM AI Engine

4-provider cascade: Groq → HuggingFace → OpenRouter → Google Gemini
Natural language → JSON workflows with auto-retry
Smart entity extraction + context-aware clarification
Self-healing AI: auto-validates & fixes malformed workflows

🧬 Dynamic Code Generation (NEW)

Intent analysis: detects when no built-in tool can handle a request
AI writes Python code: LLM generates sandboxed functions for custom capabilities
Learned tools: successfully generated code is saved to Firestore and reused for identical future requests — no LLM call needed
Secure sandbox: restricted imports, execution timeout, cross-platform (SIGALRM on Linux, threading on Windows)
Known sources: optimized RSS/scraping patterns for 10+ popular news sites and web targets
Supports: RSS feeds, web scraping, public APIs, data processing, and any task expressible in Python

🔐 Google OAuth Integration

Sign in with Google — single sign-on with full API access
Gmail — Send emails via user's own Gmail account
Google Sheets — Read, write, append data (auto-creates spreadsheets)
Google Calendar — Create events, list upcoming events
Google Drive — Upload files to user's Drive
Per-user OAuth tokens stored securely in Firestore

🔌 MCP Server (Model Context Protocol)

Streamable HTTP transport at POST /mcp
17+ tools dynamically discoverable by any MCP client
Compatible with Claude Desktop, Cursor, Windsurf, and all MCP clients
ContextMemory + execution logging for every tool call
Stateless mode — optimized for cloud deployment (Render)

📋 Tool Registry (30+ Tools + Unlimited Dynamic)

Centralized registry with JSON Schema definitions
8 categories: data_fetch, web_scraping, data_transform, notification, action, control_flow, google_oauth
Dynamic handler linking at startup
MCP-ready with inputSchema for each tool
Dynamic tools: unlimited additional capabilities via AI code generation

📊 Real-Time Data Integrations

Stock Prices: Yahoo Finance (real-time)
Crypto: Bitcoin, Ethereum, and more
Weather: OpenWeatherMap
Web Scraping: HackerNews, GitHub, Reddit, Discord
Notifications: Email (SendGrid), Gmail, SMS (Twilio), WhatsApp, Slack, Discord

⚡ Production-Ready Architecture

Workflow Validator: validates steps before execution
Context Memory: passes data between workflow steps
Execution Logger: full state tracking with Firebase
Auto port-conflict resolution: handles EADDRINUSE gracefully
Cron Scheduling for recurring automations
PostgreSQL for data persistence
Graceful shutdown handling

🎨 Beautiful Terminal UI

Hacker/terminal dark theme
Responsive design
Real-time execution status
Framer Motion animations

🛠️ Tech Stack

Frontend:

Next.js 14 (App Router) + React + TailwindCSS + Framer Motion

Backend (Node.js):

Express.js + PostgreSQL + Firebase/Firestore + Node-cron + MCP SDK

Backend (Python AI):

FastAPI + Multi-LLM (Groq, HuggingFace, OpenRouter, Gemini)
Dynamic code generation + sandboxed execution + Firebase learned tool storage

Google APIs:

OAuth2 (Sign-in + API access) · Gmail API · Sheets API · Calendar API · Drive API

MCP Server:

@modelcontextprotocol/sdk (Streamable HTTP, stateless)
17+ MCP-exposable tools with JSON Schema
ContextMemory + ExecutionLogger integration

Integrations:

SendGrid (email) · Twilio (SMS/WhatsApp) · Yahoo Finance (stocks) · OpenWeatherMap (weather) · GitHub API · HackerNews API · Reddit · Discord · Slack · Google APIs

🏗️ Architecture

graph TD
    User[User Input] --> NextUI[Next.js UI]
    MCPClient[MCP Clients] --> MCPEndpoint[POST /mcp]
    NextUI --> NodeAPI[Node.js API :3000]
    MCPEndpoint --> NodeAPI
    
    subgraph Backend
    NodeAPI --> ToolReg[Tool Registry - 30+ Tools]
    NodeAPI --> DB[(PostgreSQL)]
    NodeAPI --> Firebase[(Firebase/Firestore)]
    NodeAPI --> Scheduler[Node-Cron Scheduler]
    NodeAPI --> PyService[Python AI Service]
    NodeAPI --> MCP[MCP Server - Streamable HTTP]
    NodeAPI --> GoogleOAuth[Google OAuth2]
    MCP --> ToolReg
    end
    
    subgraph AI_Service[AI Service - Multi-LLM]
    PyService --> Groq[Groq - Primary]
    PyService --> HuggingFace[HuggingFace]
    PyService --> OpenRouter[OpenRouter]
    PyService --> Gemini[Google Gemini]
    PyService --> DynGen[Dynamic Code Generator]
    DynGen --> Sandbox[Secure Python Sandbox]
    DynGen --> LearnedTools[(Firestore - Learned Tools)]
    end
    
    subgraph Google[Google APIs]
    GoogleOAuth --> Gmail[Gmail API]
    GoogleOAuth --> Sheets[Sheets API]
    GoogleOAuth --> Calendar[Calendar API]
    GoogleOAuth --> Drive[Drive API]
    end
    
    subgraph Integrations
    ToolReg --> Stocks[Yahoo Finance]
    ToolReg --> Weather[OpenWeatherMap]
    ToolReg --> Email[SendGrid]
    ToolReg --> SMS[Twilio SMS]
    ToolReg --> Scraper[Web Scraper]
    end

🏃 Quick Start

Prerequisites

Node.js 18+
Python 3.9+
PostgreSQL
Firebase project (Firestore + service account key)
Google Cloud project (OAuth2 credentials)
API keys: At least one of Groq/HuggingFace/OpenRouter/Gemini

Installation

# 1. Clone repository
git clone https://github.com/jayyycodes/workflow-automation.git
cd workflow-automation

# 2. Install backend dependencies
npm install

# 3. Install Python AI service dependencies
cd engine-py && pip install -r requirements.txt && cd ..

# 4. Install frontend dependencies
cd Frontend/my-app && npm install && cd ../..

# 5. Set up environment variables
cp .env.example .env
# Edit .env with your API keys (see below)

# 6. Add Firebase service account
# Place your serviceAccountKey.json in the root directory

# 7. Start all services (3 terminals)
npm run dev                                          # Backend (port 3000)
cd engine-py && uvicorn app:app --reload --port 8000 # AI service (port 8000)
cd Frontend/my-app && npm run dev                    # Frontend (port 3001)

Environment Variables

# Required
JWT_SECRET=your-secret-key

# Database
DB_HOST=your-db-host
DB_PORT=5432
DB_USER=your-db-user
DB_PASSWORD=your-db-password
DB_NAME=workflow_automation

# AI Providers (at least one required)
GROQ_API_KEY=your-groq-key          # Primary (fast & free)
HUGGINGFACE_API_KEY=your-hf-key     # Fallback #1
OPENROUTER_API_KEY=your-router-key  # Fallback #2
GEMINI_API_KEY=your-gemini-key      # Fallback #3

# Google OAuth (required for Google integrations)
GOOGLE_CLIENT_ID=your-google-client-id
GOOGLE_CLIENT_SECRET=your-google-client-secret
GOOGLE_REDIRECT_URI=http://localhost:3000/auth/google/callback
FRONTEND_URL=http://localhost:3001

# Notifications
SENDGRID_API_KEY=your-sendgrid-key
EMAIL_FROM=your-email@example.com
TWILIO_ACCOUNT_SID=your-twilio-sid
TWILIO_AUTH_TOKEN=your-twilio-token
TWILIO_PHONE_NUMBER=+1234567890

# Optional
WEATHER_API_KEY=your-openweather-key
DISCORD_WEBHOOK_URL=your-discord-webhook

# Dynamic Code Generation
DYNAMIC_FEATURES_ENABLED=true
SANDBOX_MODE=local                   # 'local' or 'lambda'

Google Cloud Setup

Go to Google Cloud Console
Create a new project or select existing
Enable APIs: Gmail, Sheets, Calendar, Drive
Go to Credentials → Create OAuth 2.0 Client ID
Add authorized redirect URIs:
- Local: http://localhost:3000/auth/google/callback
- Production: https://your-backend.onrender.com/auth/google/callback
Copy Client ID and Secret to .env

Firebase Setup

Go to Firebase Console
Create a new project
Enable Firestore Database
Go to Project Settings → Service Accounts → Generate new private key
Save as serviceAccountKey.json in the root directory

🔌 MCP Server

The backend exposes a fully MCP-compatible endpoint at POST /mcp. Any MCP client can dynamically discover and execute tools.

Quick Test

# Discover all available tools
curl -X POST http://localhost:3000/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

# Execute a tool
curl -X POST http://localhost:3000/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"fetch_weather","arguments":{"city":"Mumbai"}}}'

🔐 Google Integrations

Tool	Description
`send_gmail`	Send email via user's Gmail account
`read_google_sheet`	Read data from Google Sheets
`write_google_sheet`	Write data to Sheets (auto-creates if no ID)
`append_google_sheet`	Append rows to Sheets (auto-creates if no ID)
`create_calendar_event`	Create Google Calendar event
`list_calendar_events`	List upcoming calendar events
`upload_to_drive`	Upload file to Google Drive
`list_drive_files`	List files in Google Drive

🌟 Example Automations

📧 Gmail + Crypto

"Fetch Bitcoin price and send me the result via Gmail to me@example.com"

📊 Sheets + Stock Monitoring

"Fetch AAPL stock price every 5 minutes and save to Google Sheets"

📅 Calendar Events

"Create a calendar event called Team Standup for tomorrow at 10am"

📰 HackerNews Digest

"Email me top HackerNews stories about AI every evening"

🌤️ Weather Alerts

"Send me Mumbai weather every morning at 8 AM via WhatsApp"

🧬 Dynamic — News Scraping (AI-generated code)

"Fetch top stories from Times of India and email me a summary"
"Fetch latest NDTV headlines and email me"

🧬 Dynamic — Web Scraping (AI-generated code)

"Scrape the top trending repositories on GitHub and email me"

📦 Project Structure

workflow-automation/
├── src/
│   ├── index.js                    # Express server + MCP init + auto port recovery
│   ├── mcp/
│   │   ├── mcpServer.js            # MCP server factory
│   │   └── mcpRoutes.js            # POST /mcp endpoint
│   ├── registry/
│   │   ├── toolRegistry.js         # Centralized tool registry (30+ tools)
│   │   ├── toolDefinitions.json    # JSON Schema definitions
│   │   └── mcpToolAdapter.js       # MCP ↔ Registry bridge
│   ├── automations/
│   │   ├── workflowExecutor.js     # Sequential step executor
│   │   └── stepRegistry.js         # Step handler implementations
│   ├── integrations/
│   │   ├── gmail/                  # Gmail OAuth handler
│   │   ├── sheets/                 # Google Sheets handler (auto-create)
│   │   ├── calendar/               # Google Calendar handler (smart dates)
│   │   └── drive/                  # Google Drive handler
│   ├── services/
│   │   └── googleOAuth.js          # OAuth2 flow + token management
│   ├── config/
│   │   └── firebase.js             # Firebase/Firestore config
│   ├── utils/
│   │   ├── executionLogger.js      # State tracking + ContextMemory
│   │   └── logger.js               # Winston logger
│   └── routes/                     # REST API routes
├── engine-py/                      # Python AI service (FastAPI)
│   ├── app.py                      # Multi-LLM provider cascade + /execute_dynamic
│   ├── config.py                   # LLM configuration
│   ├── prompts.py                  # System prompts + dynamic code generation prompts
│   ├── dynamic_resolver.py         # Capability gap resolver + Firestore learned tools
│   ├── sandbox.py                  # Secure sandbox for AI-generated code execution
│   └── validator.py                # Workflow validation
├── Frontend/my-app/                # Next.js 14 frontend
├── serviceAccountKey.json          # Firebase service account (gitignored)
├── render.yaml                     # Render deployment config
└── .env                            # Environment variables (gitignored)

🔥 Production Deployment

Deploy to Render

Push to GitHub:

git add .
git commit -m "production deployment"
git push origin main

Set Environment Variables in Render dashboard for both services:
- Backend (Node.js): All .env variables + production Google OAuth credentials
- Python AI: GROQ_API_KEY, HUGGINGFACE_API_KEY, OPENROUTER_API_KEY, GEMINI_API_KEY
Google Cloud Console: Add your Render backend URL as authorized redirect URI:
```
https://your-backend.onrender.com/auth/google/callback
```
Deploy — Render auto-deploys on git push!

Deploy Frontend to Vercel

Import your GitHub repo in Vercel
Set root directory to Frontend/my-app
Add environment variable: NEXT_PUBLIC_API_URL=https://your-backend.onrender.com
Deploy!

🤝 Contact & Collaboration

This project is proprietary and not open for public contributions at this time.

For business inquiries, partnerships, or collaboration:

GitHub: @jayyycodes

📄 License

This source code is publicly visible for evaluation and portfolio purposes only. Commercial use, redistribution, or derivative works require explicit written permission from the author.

🙏 Acknowledgments

Groq — Primary AI engine (fast inference)
Google Gemini — Fallback AI engine
Anthropic MCP — Model Context Protocol standard
OpenRouter — Multi-model routing
Firebase — User token storage, execution logging & learned tool storage
Render — Backend deployment
Vercel — Frontend hosting

⚡ Transform plain English into powerful automations ⚡

🔌 MCP-compatible · 30+ tools · Dynamic Code Gen · Multi-LLM · Google Integrated · Production-ready

Built with ❤️ by @jayyycodes

Name		Name	Last commit message	Last commit date
Latest commit History 39 Commits
Frontend/my-app		Frontend/my-app
docs		docs
engine-py		engine-py
images		images
lambda		lambda
src		src
.env.example		.env.example
.gitignore		.gitignore
LICENSE		LICENSE
Prototype Development Submission _ AWS AI for Bharat Hackathon.pptx		Prototype Development Submission _ AWS AI for Bharat Hackathon.pptx
README.md		README.md
design.md		design.md
example-prompts.txt		example-prompts.txt
package-lock.json		package-lock.json
package.json		package.json
render.yaml		render.yaml
requirements.md		requirements.md
runtime.txt		runtime.txt

Folders and files

Latest commit

History

Repository files navigation

🤖 Smart Workflow Automation

✨ What Is This?

Just say:

🎯 Live Demo

🚀 Features

🤖 Multi-LLM AI Engine

🧬 Dynamic Code Generation (NEW)

🔐 Google OAuth Integration

🔌 MCP Server (Model Context Protocol)

📋 Tool Registry (30+ Tools + Unlimited Dynamic)

📊 Real-Time Data Integrations

⚡ Production-Ready Architecture

🎨 Beautiful Terminal UI

🛠️ Tech Stack

🏗️ Architecture

🏃 Quick Start

Prerequisites

Installation

Environment Variables

Google Cloud Setup

Firebase Setup

🔌 MCP Server

Quick Test

🔐 Google Integrations

🌟 Example Automations

📧 Gmail + Crypto

📊 Sheets + Stock Monitoring

📅 Calendar Events

📰 HackerNews Digest

🌤️ Weather Alerts

🧬 Dynamic — News Scraping (AI-generated code)

🧬 Dynamic — Web Scraping (AI-generated code)

📦 Project Structure

🔥 Production Deployment

Deploy to Render

Deploy Frontend to Vercel

🤝 Contact & Collaboration

📄 License

🙏 Acknowledgments

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages