Agent Context Protocols Enhance Collective Inference

2025-05-20

AI Summary

Existing multi-agent systems rely on unstructured natural language communication, resulting in ambiguous collaboration, poor fault tolerance, and limited cross-domain interoperability. This paper introduces Agent Context Protocols (ACPs)—a domain-agnostic, structured collaboration framework comprising: (1) persistent execution blueprints modeled as directed acyclic graphs (DAGs) to explicitly capture intermediate-result dependencies; (2) standardized message schemas and built-in fault-recovery mechanisms to ensure robust inter-agent communication; and (3) modular protocol extension interfaces enabling seamless integration of heterogeneous, domain-specialized agents. ACPs are the first framework to unify dependency modeling with protocol standardization, significantly enhancing collaborative reasoning reliability. Evaluated on AssistantBench’s long-horizon web tasks, ACP-based agents achieve 28.3% task accuracy. Moreover, the generated multimodal technical reports surpass leading commercial AI systems in quality, ranking first in human evaluation.

Technology Category

• Multiagent Systems: Agent Communication
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

AI agents have become increasingly adept at complex tasks such as coding, reasoning, and multimodal understanding. However, building generalist systems requires moving beyond individual agents to collective inference -- a paradigm where multi-agent systems with diverse, task-specialized agents complement one another through structured communication and collaboration. Today, coordination is usually handled with imprecise, ad-hoc natural language, which limits complex interaction and hinders interoperability with domain-specific agents. We introduce Agent context protocols (ACPs): a domain- and agent-agnostic family of structured protocols for agent-agent communication, coordination, and error handling. ACPs combine (i) persistent execution blueprints -- explicit dependency graphs that store intermediate agent outputs -- with (ii) standardized message schemas, enabling robust and fault-tolerant multi-agent collective inference. ACP-powered generalist systems reach state-of-the-art performance: 28.3 % accuracy on AssistantBench for long-horizon web assistance and best-in-class multimodal technical reports, outperforming commercial AI systems in human evaluation. ACPs are highly modular and extensible, allowing practitioners to build top-tier generalist agents quickly.

Problem

• Enhancing multi-agent communication for collective inference
• Replacing ad-hoc language with structured agent protocols
• Improving interoperability in task-specialized AI systems

Innovation

• Agent context protocols for structured agent communication
• Persistent execution blueprints with dependency graphs
• Standardized message schemas for robust coordination