This repository serves as a curated collection of research papers and resources focused on "Agentic Web," exploring the evolution of the web towards AI-driven agents. It targets researchers and developers in AI, particularly those interested in web automation, information retrieval, recommendation systems, multi-agent learning, and AI safety. The primary benefit is a centralized hub for understanding the state-of-the-art and ongoing research in this rapidly developing field.

How It Works

The project acts as a living bibliography, categorizing and listing academic papers related to agentic web technologies. It covers diverse sub-fields such as agent planning, multi-agent learning, and safety/security considerations for AI agents interacting with the web. The organization by research area allows users to quickly navigate and identify relevant studies for their specific interests.

Quick Start & Requirements

This repository is a collection of research papers and does not have a direct installation or execution command. It requires no specific software prerequisites beyond a web browser to view the listed papers.

Highlighted Details

• Comprehensive categorization of research papers across key agentic web domains.
• Includes foundational and recent publications, spanning from 1972 to projected 2025 works.
• Features a dedicated section for "Safety and Security" concerning AI agents.
• Provides a "Benchmark" section for evaluating agent capabilities.

Maintenance & Community

The repository is marked as "under actively development" and welcomes contributions via forking, issue reporting, or direct email to a listed corresponding author. It aims to be a community-driven resource for tracking agentic web research.

Licensing & Compatibility

The repository itself does not specify a license. The content consists of links to research papers, which are subject to their respective publisher's licenses and copyright.

Limitations & Caveats

As a curated list of papers, this repository does not provide any code, tools, or executable agents. Its value is purely informational, relying on the community for updates and accuracy. The "2025" dates on many papers suggest they may be pre-prints or future work not yet published.