Workflow Generator

AI-powered workflow blueprint generator using Gemini AI and Model Context Protocol (MCP).

Overview

This system generates structured workflow blueprints from natural language descriptions. It provides a modern web interface for creating, visualizing, and exporting workflows in multiple formats (JSON, YAML, BPMN).

Features

• 🧠 AI-Powered Generation: Converts text descriptions into structured workflows
• 🎨 Modern Web Interface: Clean, intuitive UI for workflow creation
• 📤 Multiple Export Formats: JSON, YAML, and BPMN support
• 🔄 Real-time Processing: Instant workflow generation and preview
• 🛠️ MCP Integration: Built on Model Context Protocol for extensibility

Prerequisites

For Docker:

• Docker and Docker Compose installed
• Gemini API key (get one at Google AI Studio)

For Local Python:

• Python 3.11 or higher
• Gemini API key (get one at Google AI Studio)

Setup

1. Install Dependencies

   pip install -r requirements.txt

2. Configure API Key

   Create a .env file in the project root:

   GEMINI_API_KEY=your_api_key_here
   

3. Verify Configuration

   Ensure config/data.json exists with required settings.

Running

Option 1: Docker (Recommended)

Build the image first:

docker build -t workflow-generator .

Using Docker Compose (modern Docker/cloud environments):

docker compose up -d

Or legacy docker-compose (if above doesn't work):

docker-compose up -d

Alternative: Using docker run (for cloud environments without compose):

docker run -d \
  --name workflow-generator \
  -p 5000:5000 \
  -e GEMINI_API_KEY=your_api_key_here \
  -e GEMINI_MODEL=gemini-2.5-flash \
  -v $(pwd)/data:/app/data \
  -v $(pwd)/client:/app/client \
  -v $(pwd)/server:/app/server \
  --restart unless-stopped \
  workflow-generator

Then open your browser to: http://localhost:5000

Stop the application:

# If using compose:
docker compose down

# If using docker run:
docker stop workflow-generator
docker rm workflow-generator

View logs:

# If using compose:
docker compose logs -f

# If using docker run:
docker logs -f workflow-generator

Rebuild after changes:

# If using compose:
docker compose up --build -d

# If using docker run:
docker stop workflow-generator
docker rm workflow-generator
docker build -t workflow-generator .
# Then run the docker run command again

Option 2: Local Python

Start the application:

start.bat

Or manually:

python server\api_server.py

Then open your browser to: http://localhost:5000

File Structure

neu/
├── server/                # Server components
│   ├── server.py         # MCP server backend
│   ├── api_server.py     # REST API server
│   └── server_web.py     # Static file server
├── client/                # Frontend files
│   ├── index.html        # Web interface
│   ├── script.js         # Frontend logic
│   └── styles.css        # Styles
├── config/
│   └── data.json         # Configuration
├── data/
│   └── workflows/        # Generated workflows
├── start.bat             # Startup script
└── requirements.txt      # Python dependencies

Usage

1. Launch the application using start.bat
2. Open the web interface at http://localhost:5000
3. Enter a workflow description in natural language
4. Select the domain (HR, IT, Customer Service, etc.)
5. Generate and view your workflow
6. Export in your preferred format (JSON, YAML, BPMN)

Example Prompts

"Create an employee onboarding workflow with HR approval and IT setup steps"

"Design a customer support ticket escalation process with SLA checks"

"Build a purchase approval workflow with budget validation and multi-level approval"

Architecture

• MCP Server (server.py): Handles workflow generation logic using Gemini AI
• REST API (api_server.py): Bridges web interface and MCP server
• Web Frontend: Modern HTML/JS interface for workflow creation

API Key

The system requires a Google Gemini API key:

1. Visit Google AI Studio
2. Create an API key
3. Add it to your .env file

Output

Generated workflows are saved to data/workflows/ in your selected format:

• workflow_name.json - JSON format
• workflow_name.yaml - YAML format
• workflow_name.bpmn - BPMN XML format

Troubleshooting

"GEMINI_API_KEY not found"

• Ensure .env file exists in project root
• Verify API key is correctly set

"Port 5000 already in use"

• Stop other applications using port 5000
• Or edit api_server.py to use a different port

"Module not found"

• Run pip install -r requirements.txt

License

MIT License