PEEK: Context Map as an Orientation Cache for Long-Context LLM Agents

TL;DR

PEEK introduces a reusable context map system that enhances long-context LLM agents' ability to efficiently and accurately interact with recurring external contexts by maintaining orientation knowledge.

Contribution

The paper presents PEEK, a novel system that caches and maintains orientation knowledge as a context map, improving long-context reasoning and context learning in LLM agents.

Findings

PEEK improves reasoning and aggregation by 6.3-34.0% over baselines.

PEEK reduces iterations by 93-145 and costs by 1.7-5.8x compared to ACE.

PEEK enhances solving rate and rubric accuracy by 6.0-14.0% and 7.8-12.1%, respectively.

Abstract

Large language model (LLM) agents increasingly operate over long and recurring external contexts, like document corpora and code repositories. Across invocations, existing approaches preserve either the agent's trajectory, passive access to raw material, or task-level strategies. None of them preserves what we argue is most needed for repeated same-context workloads: reusable orientation knowledge (e.g., what the context contains, how it is organized, and which entities, constants, and schemas have historically been useful) about the recurring context itself. We introduce PEEK, a system that caches and maintains this orientation knowledge as a context map: a small, constant-sized artifact in the agent's prompt that gives it a persistent peek into the external context. The map is maintained by a programmable cache policy with three modules: a Distiller that extracts transferable knowledge from inference-time signals, a Cartographer that translates it into structured edits, and a priority-based Evictor that enforces a fixed token budget. On long-context reasoning and information aggregation, PEEK improves over strong baselines by 6.3-34.0% while using 93-145 fewer iterations and incurring 1.7-5.8x lower cost than the state-of-the-art prompt-learning framework, ACE. On context learning, PEEK improves solving rate and rubric accuracy by 6.0-14.0% and 7.8-12.1%, respectively, at 1.4x lower cost than ACE. These gains generalize across LMs and agent architectures, including OpenAI Codex, a production-grade coding agent. Together, these results show that a context map helps long-context LLM agents interact with recurring external contexts more accurately and efficiently.