Navigator AI empowers users and developers to seamlessly automate tasks within web browsers (with app support coming in the future). Unlike traditional browser automation tools, Navigator AI will offer:

• A component library for direct integration into web applications
• A browser extension for automating tasks in your browser

Consider it as Cursor/Windsurf for websites and applications. Big shoutout to Browser-Use as Navigator AI is inspired by them and currently is a kind of Chrome extension version that does what they do inside but inside YOUR browser while you retain full control to pause, stop, or restart.

⭐️ P.s. Drop a star if you want a completely FREE and open-source alternative to Manus AI. I aim to make Navigator AI on par with Manus AI. Your support will motivate me a lot. Thank you!

• Direct Web Integration: Embeddable React/framework components allow developers to add browser automation capabilities directly within their web/mobile applications. Users can easily use voice/text to automatically execute workflows on your app.
• User-Friendly Extension: A browser extension that allows users to create, manage, and run repeatable workflows directly in their browser.
• Knowledge Base Integration: Add custom rules, documentation, and knowledge bases that the agent will prioritize over its LLM-based workflow, making the agent specific to YOUR application.
• Self-Improvement: The agent improves over time based on how you interact with different pages, especially when not actively using the agent.

• Frontend:
  • React, Vite, TypeScript (for both component library and extension)
  • Packaged in a Turborepo for efficient management
• Backend:
  • Python, FastAPI
• Database:
  • PostgreSQL (primary data storage)
  • Redis (caching)
  • Weaviate (vector database, deployed via Docker)

• Node.js (v16+) - Install Guide
• Package manager:
  • pnpm - Install Guide (npm install -g pnpm)
  • OR npm (comes with Node.js)
• Python 3.9+ - Install Guide
• Poetry (Python dependency management) - Install Guide
• Docker and Docker Compose (for database services) - Install Guide

1. Clone the repository

   git clone https://github.com/SohamRatnaparkhi/navigator-ai.git
   cd navigator-ai

2. Install dependencies

   # Install Python dependencies
   cd apps/server
   poetry install
   
   # Install Node dependencies
   cd ../extension
   pnpm install
   # OR
   npm install

3. Run the development server

   pnpm run dev:server
   # OR
   npm run dev:server

4. Run Redis

   cd apps/server
   docker compose up -d

5. Build and install the extension

   # Build the extension
   pnpm run build
   # OR
   npm run build

   Then:

   • Open Chrome and navigate to chrome://extensions
   • Enable "Developer mode" (toggle in the top-right corner)
   • Click "Load unpacked" and select the /apps/extension/dist directory
   • The extension should now appear in your browser toolbar

• Deep agentic workflows

  • Planner at the top/each step
  • Support for multiple actions like switching tabs and copy-pasting

• Visual Task Builder (Extension)

  • Develop a drag-and-drop interface for creating automation workflows
  • Add support for conditional logic and branching

• Advanced DOM Interaction

  • Implement sophisticated element selection methods using vision LLMs
  • Add support for handling dynamic content

• Self-Improving Agents

  • Implement feedback loops to learn from user corrections
  • Track user activity patterns (with permission) to improve automation
  • Develop metrics for measuring and reporting agent improvement

• Knowledge Base Enhancement

  • Create an interface for managing custom rules and documentation
  • Implement priority weighting for different knowledge sources
  • Add support for importing existing documentation

• Third-party Integrations

  • Website-specific integrations (AWS, GCP, Amazon, etc.)
  • Multiple LLM provider support
  • API connections to popular services

• Complex Web Interactions

  • Support for iframes and shadow DOM
  • Handling authentication and user sessions
  • Intelligent error recovery and pause mechanisms

• Workflow Management

  • Record and replay functionality for capturing user workflows
  • Scheduled tasks with time/interval specifications
  • Workflow sharing and importing capabilities

• Notification System

  • Alert users when automation encounters obstacles
  • Provide detailed reporting on automation performance
  • Suggest improvements based on execution patterns

Contributions are welcome! This project has areas for improvement and we appreciate your help. Please feel free to submit a Pull Request or create an Issue if you find a bug.

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