The Context Gathering Decision Process: A POMDP Framework for Agentic Search

TL;DR

This paper introduces a POMDP-based framework called the Context Gathering Decision Process (CGDP) for improving agentic search in large environments by refining belief states and halting unproductive searches.

Contribution

It formalizes the CGDP for LLM agents, models behavior as approximate Thompson Sampling, and proposes modular interventions that enhance reasoning and reduce token usage.

Findings

Replacing implicit state with CGDP belief state improves multi-hop reasoning by up to 11.4%.

Programmatic exhaustion gate reduces token usage by up to 39% without performance loss.

Framework guides modular improvements to agentic search in complex environments.

Abstract

Large Language Model (LLM) agents are deployed in complex environments -- such as massive codebases, enterprise databases, and conversational histories -- where the relevant state far exceeds their context windows. To navigate these spaces, an agent must iteratively explore the environment to find relevant information. However, without explicit infrastructure, an agent's working memory can degrade into lossy representations of the search state, resulting in redundant work (e.g. repetitive looping) and premature stopping. In this work, we formalize this challenge as the Context Gathering Decision Process (CGDP), a specialized Partially Observable Markov Decision Process, where an agent's objective is to adaptively refine its belief state to isolate the necessary information for a task. We model an LLM's behavior as approximate Thompson Sampling within this CGDP, and introduce a predicate-based method that decomposes an LLM's implicit search into explicit and modular operations. We then derive two plug-and-play interventions for iterative LLM agents: a persistent, predicate-based belief state that bounds context while preserving multi-hop reasoning, and a programmatic exhaustion gate that halts unproductive search without premature stopping. Across four methods and three question-answering domains, we empirically validate that replacing an LLM's implicit state with our CGDP-motivated belief state improves multi-hop reasoning by up to $11.4\%$; while the modular programmatic exhaustion detection saves up to $39\%$ of tokens without any degradation in agent performance. Ultimately, we argue that framing the LLM agent loop as a CGDP can guide the design of modular, non-interfering improvements to agentic search harnesses.