SearchOS-V1: Towards Robust Open-Domain Information-Seeking Agent Collaboration

2026-07-16

AI Summary

This work addresses the challenge of open-domain information-seeking agents repeatedly querying due to difficulties in tracking long interaction histories, which leads to wasted resources and degraded performance. The authors formulate the task as relation schema completion grounded in cited evidence and introduce a multi-agent collaborative framework featuring a Search-Oriented Context Management mechanism that explicitly maintains a shared, persistent search state. This mechanism externalizes search progress through Frontier Tasks, an evidence graph, a coverage graph, and failure memory, enabling efficient state management and scheduling. Integrated with a search middleware and a hierarchical skill system, the approach effectively avoids redundant failures. Evaluated on the WideSearch and GISA benchmarks, the proposed method significantly outperforms existing single- and multi-agent approaches, achieving state-of-the-art results across all metrics.

Abstract

Recent advances in Tool-Integrated Large Language Models have made web search a core capability of information-seeking agents. However, as interaction histories grow, agents increasingly struggle to track task progress. When search attempts fail to yield useful evidence, current single- and multi-agent systems can become trapped in repetitive loops, wasting search budgets and ultimately compromising the quality and completeness of the final output. We introduce SearchOS, a system-level multi-agent framework that turns fragile, implicit search progress into explicit, persistent, and shared state. First, we formulate open-domain information seeking as relational schema completion with grounded citations, where agents discover entities, populate attributes across linked tables, and anchor each value to source evidence. Then we design Search-Oriented Context Management (SOCM), which externalizes the evolving state into Frontier Task, an Evidence Graph, a Coverage Map, and Failure Memory. Built on SOCM, SearchOS applies a pipeline-parallel scheduling mechanism that overlaps the execution of sub-agents and continuously refills freed slots with tasks targeting unresolved coverage gaps to improve utilization and throughput. To schedule and control the execution of search agents, SearchOS introduces a Search Tool Middleware Harness that intercepts model and tool interactions to record grounded evidence and react to stalls or budget exhaustion, and provides a reusable hierarchical skill system comprising strategy and access skills to augment the agents' search process and avoid repeating failed search patterns across runs. On WideSearch and GISA, SearchOS leads all metrics among the evaluated single- and multi-agent baselines, paving the way toward robust information-seeking collaboration.

Problem

• open-domain information seeking
• agent collaboration
• search progress tracking
• repetitive search loops
• search budget efficiency

Innovation

• SearchOS
• multi-agent collaboration
• relational schema completion
• search-oriented context management
• pipeline-parallel scheduling