Agent-Centric Web Protocols: Redesigning the Internet for Autonomous Collaboration

Inspired by the Internet of Agents concept (Chen et al., 2024) and the DOVIS/AgentRank architecture (Krishnamachari & Rajesh, 2025), this research proposes a fundamental shift: treat autonomous agents—not just humans or devices—as first-class citizens in protocol design. Current protocols (HTTP, MQTT, etc.) are not optimized for the needs of agent collaboration, such as dynamic trust evaluation, agent discovery, negotiation, and secure delegation. This project would design a protocol suite with native support for agent attributes (capabilities, provenance, reputation), secure agent-to-agent messaging, incentive mechanisms, and privacy-by-design. By embedding performance-and-trust-aware routing and orchestration at the protocol level, we move beyond stateless transactions to a fabric that supports rich, persistent, and trustworthy agent ecosystems. Such a protocol architecture could unlock the full potential of LLM-based agents and multi-agent AI systems, laying the groundwork for the next evolution of the web.

References:

1. Internet of Agents: Weaving a Web of Heterogeneous Agents for Collaborative Intelligence. Weize Chen, Ziming You, Ran Li, Yitong Guan, Cheng Qian, Chenyang Zhao, Cheng Yang, Ruobing Xie, Zhiyuan Liu, Maosong Sun (2024). International Conference on Learning Representations.
2. Internet 3.0: Architecture for a Web-of-Agents with it's Algorithm for Ranking Agents. Rajesh T. Krishnamachari, Srividya Rajesh (2025). arXiv.org.