FastAPI backend for a conversational RAG agent that embeds documents into Qdrant, runs semantic and hybrid search, and answers questions with LLMs.
hrithikkoduri/WebRover
An open-source autonomous AI web agent that uses LLMs and Playwright to browse sites, complete tasks, and run multi-step research.
- Stars
- 1,025
- Forks
- 179
- Open issues
- 12
- Last commit
- 18 days ago
What is it?
WebRover is an AI web agent that takes a natural-language request, decides whether it is a browsing task or a research question, and carries it out by interacting with web pages. It is built with Python and FastAPI on the backend, LangGraph for agent state management, and Playwright for browser automation, with a Next.js and TypeScript chat interface on the frontend.
The project offers three agent types: a Task Agent that plans and executes multi-step actions on websites, a Research Agent that gathers and summarizes information from sources, and a Deep Research Agent that breaks a topic into subtopics and produces a structured paper with citations. A vector store and RAG pipeline support retaining and retrieving collected information. Supported LLMs include GPT-4o, o3-mini and Claude 3.5 Sonnet.
The browser runs locally under the user's control, and results can be returned in chat, downloaded as PDF, or exported to Google Docs.
Key features
- 01Three agent types for task automation, research, and deep research
- 02Local browser control via Playwright with multi-tab management
- 03Streaming of agent actions and reasoning in a chat interface
- 04RAG pipeline with a vector store for retaining collected information
- 05Export of results to PDF, Google Docs, or clipboard
Use cases
- Automating repetitive multi-step tasks on websites
- Gathering and summarizing information from multiple web sources
- Drafting research papers with citations on a given topic
Quick start
- 1
Clone the repository and enter the backend folder
git clone https://github.com/hrithikkoduri18/WebRover.git - 2
Install dependencies with Poetry (after setting Python 3.12 and running poetry install)
poetry install - 3
Start the backend API server on port 8000
uvicorn app.main:app --reload --port 8000 - 4
Install frontend dependencies and run the frontend dev server (in the frontend folder)
npm install
Things to know
- Requires an OpenAI or Anthropic API key to run the LLM-based agents.
- Setup is multi-step, needing Poetry, Python 3.12, and a separate Next.js frontend.
- Some Mac users may hit browser connection issues that need manual workarounds.
In collections
Topics
FAQ
- What is WebRover?
- An open-source autonomous AI web agent that uses LLMs and Playwright to browse sites, complete tasks, and run multi-step research. WebRover is an AI web agent that takes a natural-language request, decides whether it is a browsing task or a research question, and carries it out by interacting with web pages. It is built with Python and FastAPI on the backend, LangGraph for agent state management, and Playwright for browser automation, with a Next.js and TypeScript chat interface on the frontend.
- Is WebRover free to use? What license does it have?
- WebRover is open source under the MIT license. Check the license terms before using it in commercial products.
- Is WebRover still maintained?
- Actively maintained. The last commit to WebRover was on Oct 19, 2026.
- What can I use WebRover for?
- Automating repetitive multi-step tasks on websites Gathering and summarizing information from multiple web sources Drafting research papers with citations on a given topic
- How do I install WebRover?
- Clone the repository and enter the backend folder: git clone https://github.com/hrithikkoduri18/WebRover.git Install dependencies with Poetry (after setting Python 3.12 and running poetry install): poetry install Start the backend API server on port 8000: uvicorn app.main:app --reload --port 8000 Install frontend dependencies and run the frontend dev server (in the frontend folder): npm install
- Where can I get WebRover?
- The source code and installation instructions are on GitHub: https://github.com/hrithikkoduri/WebRover
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
An open-source autonomous research agent that plans questions, gathers web and local sources, and writes cited reports with any LLM provider.
Open-source long-horizon super agent harness that orchestrates sub-agents, memory, and sandboxes to research, code, and create across tasks lasting minutes to hours.
A tutorial repo of five Jupyter notebooks that walk through building a LangGraph-based deep research agent from scratch.
A deep research skill and pipeline for Claude Code and OpenAI Codex that gathers web and scholarly sources into a persistent, searchable markdown vault and writes cited reports.
Multi-agent LLM pipeline that turns a research paper or topic into a cited scorecard on technology readiness, patents, market and risks.