WebRover is an autonomous AI agent designed for web navigation, task automation, and deep research. It caters to users needing to execute web-based tasks, gather information efficiently, or conduct in-depth research and generate academic content. The system offers specialized agents for distinct use cases, providing a versatile tool for both quick actions and comprehensive analysis.

How It Works

WebRover employs a modular architecture with three specialized agents: Task, Research, and Deep Research. It utilizes LangGraph for state management, enabling complex navigation and decision-making. Playwright handles reliable web automation, including custom element detection and content extraction. A RAG pipeline with vector store integration facilitates efficient information storage and retrieval, powering LLM integrations (GPT-4o, Claude-3.5 Sonnet) for context-aware navigation and structured output generation.

Quick Start & Requirements

• Backend: Clone repo, install Poetry, set Python 3.12, activate Poetry shell, run poetry install. Set environment variables (OPENAI_API_KEY, LANGCHAIN_API_KEY, etc.) in .env. Run backend with uvicorn app.main:app --reload --port 8000.
• Frontend: In a new terminal, cd frontend, run npm install, then npm run dev.
• Prerequisites: Python 3.12, Node.js, npm, OpenAI API key, Langchain API key, Anthropic API key.
• Notes: For Mac users facing browser connection issues, try pkill -9 "Chrome" or change the websocket port in backend/Browser/webrover_browser.py.
• Demo Videos: Deep Research Agent, Task Agent

Highlighted Details

• Three specialized agents: Task, Research, and Deep Research (academic paper generation).
• Dynamic agent selection based on task complexity.
• Real-time agent state visualization and action streaming.
• Local browser instance via Playwright for privacy and control.
• Multi-source verification and academic paper generation with citations.

Maintenance & Community

• Project maintained by @hrithikkoduri.
• Contribution guidelines provided via standard Git workflow (fork, branch, commit, PR).

Licensing & Compatibility

• Licensed under the MIT License.
• Permissive license suitable for commercial use and integration with closed-source projects.

Limitations & Caveats

The project is actively developed with a new "Deep Research Agent" feature. Specific performance benchmarks or detailed comparisons to existing tools are not provided in the README.