Transform plain English into production-ready test automation. Mark 1 is an intelligent test generation platform that converts natural language descriptions into executable Robot Framework code using a sophisticated multi-agent AI system. No coding required—just describe what you want to test.

"Open Flipkart and search for shoes and then get the first product name" ↓ [4 AI Agents Working Together] ↓ ✅ Working Robot Framework Test 

• 🎯 95%+ Success Rate - Vision-based element detection that actually works
• ⚡ 3-5x Faster - Batch processing finds all elements in one session
• 🧠 Context-Aware - AI understands your workflow, not just individual steps
• 🔒 Privacy-First - Run locally with Ollama or use cloud models
• 📦 Zero Setup - One command to start, works out of the box
• 🎨 Beautiful Reports - Detailed HTML logs for easy debugging

• ✅ Python 3.9+ - Download
• ✅ Docker Desktop - Install (must be running!)
• ✅ Git - Install
• ✅ Google Gemini API Key - Get Free Key

# 1. Clone the repository git clone https://github.com/your-repo/mark-1.git cd mark-1 # 2. Configure your API key cp src/backend/.env.example src/backend/.env # Edit src/backend/.env and add your GEMINI_API_KEY # 3. Start Mark 1 chmod +x run.sh ./run.sh # 4. Start BrowserUse service (in another terminal) python tools/browser_use_service.py

1. Open http://localhost:5000 in your browser
2. Enter a test description:

   Open Flipkart and search for shoes and then get the first product name 

3. Click "Generate & Run"
4. Watch the magic happen! ✨

"Open Flipkart and search for shoes and then get the first product name" 

"Navigate to GitHub using url https://github.com/monkscode, and then get the name of the Pinned project" 

"Go to google.com and search for python tutorials" 

Pro Tip: Be specific about what you want. Mention exact elements like "first product name" or "search button in header".

Mark 1 uses a multi-agent AI system to transform your natural language into working tests:

Your Query → [AI Processing] → Robot Framework Code → Execution → Results 

The Process:

1. Intelligent Planning - Query analyzed and broken into precise steps
2. Smart Element Detection - AI finds web elements with 95%+ accuracy
3. Code Generation - Transforms steps into production-ready Robot Framework code
4. Quality Assurance - Validates code before execution
5. Isolated Execution - Runs in clean Docker containers

What You Get:

• ✅ Working test code in Robot Framework format
• ✅ Detailed HTML reports with step-by-step execution logs
• ✅ Real-time progress updates
• ✅ Validated locators that work on dynamic websites

mark-1/ ├── src/backend/ # FastAPI backend with AI agents ├── tools/ # Browser automation utilities ├── robot_tests/ # Generated tests & reports (auto-created) │ └── {run-id}/ │ ├── test.robot # Your generated test │ ├── log.html # Detailed execution log │ └── report.html # Test summary ├── docs/ # Documentation ├── run.sh # One-command startup └── README.md # You are here! 

• Library Switching Guide - Switch between Browser Library & Selenium ⭐
• Configuration Guide - Environment variables and settings
• Troubleshooting - Fix common issues
• FAQ - Frequently asked questions
• Best Practices - Get the most out of Mark 1
• Architecture - How Mark 1 works under the hood
• Contributing - Help improve Mark 1

Your Input:

Open Flipkart and search for shoes and then get the first product name 

Generated Code (Browser Library):

*** Settings *** Library Browser *** Variables *** ${browser} chromium ${headless} False *** Test Cases *** Search Shoes On Flipkart New Browser ${browser} headless=${headless} New Context viewport=None New Page https://www.flipkart.com Fill Text name=q shoes Keyboard Key press Enter ${product_name}= Get Text xpath=(//div[@class='_4rR01T'])[1] Log First product name: ${product_name} Close Browser

Result: Working test + detailed HTML report in ~20 seconds.

Note: Code format depends on your ROBOT_LIBRARY setting (browser or selenium).

Create a .env file in src/backend/:

# AI Provider MODEL_PROVIDER=online GEMINI_API_KEY=your-actual-api-key-here ONLINE_MODEL=gemini-2.5-flash # Application APP_PORT=5000 # Browser Automation BROWSER_USE_SERVICE_URL=http://localhost:4999 BROWSER_USE_TIMEOUT=900 # Robot Framework Library (selenium or browser) ROBOT_LIBRARY=browser

Get your free Gemini API key: Google AI Studio

For detailed configuration options, see the Configuration Guide.

Mark 1 supports two Robot Framework libraries for test execution:

ROBOT_LIBRARY=browser

Benefits:

• ✅ 2-3x faster test execution
• ✅ Better AI compatibility - LLMs understand JavaScript/Playwright better
• ✅ Modern web support - Shadow DOM, iframes, SPAs work seamlessly
• ✅ Auto-waiting built-in - No explicit waits needed
• ✅ Powerful locators - Text-based, role-based, and traditional selectors
• ✅ Consistent validation - Same engine (Playwright) for generation and execution

When to use: New projects, modern websites, performance-critical tests

ROBOT_LIBRARY=selenium

Benefits:

• ✅ Mature and stable - Battle-tested library
• ✅ Wide compatibility - Works with older websites
• ✅ Familiar syntax - Traditional Selenium approach

When to use: Existing projects, legacy websites, Selenium expertise

Switching is easy: Just change ROBOT_LIBRARY in your .env file and restart Mark 1!

Test the API directly:

chmod +x test.sh ./test.sh

Or use curl:

curl -X POST http://localhost:5000/generate-and-run \ -H "Content-Type: application/json" \ -d '{"query": "go to google.com and search for python tutorials"}'

Common Issues:

• "Docker is not available" - Make sure Docker Desktop is running
• "GEMINI_API_KEY not found" - Check your .env file in src/backend/
• "Port 5000 already in use" - Change APP_PORT in your .env file
• Tests fail with "Element not found" - Try being more specific in your query

For detailed troubleshooting, see the Troubleshooting Guide.

We welcome contributions! Whether it's bug fixes, new features, or documentation improvements, your help makes Mark 1 better for everyone.

Please see our Contributing Guide for detailed instructions on:

• Setting up your development environment
• Making and testing changes
• Submitting pull requests
• Code guidelines and best practices

By submitting a pull request, you agree to our Contributor License Agreement.

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

• 📚 Documentation: Check the docs folder
• 🐛 Bug Reports: GitHub Issues
• 💬 Discussions: GitHub Discussions
• 💡 Feature Requests: Open an issue with the enhancement label

If Mark 1 helps streamline your testing workflow:

• ⭐ Star this repository
• 🐛 Report issues and help us improve
• 💡 Suggest features
• 🤝 Contribute code
• 📢 Share your experience

Built with ❤️ for the test automation community

Mark 1 is not affiliated with or endorsed by Google, Robot Framework, or any mentioned websites. All trademarks belong to their respective owners.