Environment Maps: Structured Environmental Representations for Long-Horizon Agents
Yenchia Feng, Chirag Sharma, Karime Maamari

TL;DR
This paper introduces Environment Maps, a structured, persistent environmental representation that enhances long-horizon agent planning by consolidating diverse evidence into a graph, improving success rates in complex tasks.
Contribution
The paper presents Environment Maps, a novel structured environmental representation that improves long-horizon agent performance by integrating heterogeneous evidence into a persistent, editable graph.
Findings
Agents with Environment Maps achieve 28.2% success rate, nearly doubling baseline performance.
Environment Maps outperform raw trajectory data access, which yields 23.3% success.
Structured representations enable more robust, interpretable, and incremental long-horizon planning.
Abstract
Although large language models (LLMs) have advanced rapidly, robust automation of complex software workflows remains an open problem. In long-horizon settings, agents frequently suffer from cascading errors and environmental stochasticity; a single misstep in a dynamic interface can lead to task failure, resulting in hallucinations or trial-and-error. This paper introduces : a persistent, agent-agnostic representation that mitigates these failures by consolidating heterogeneous evidence, such as screen recordings and execution traces, into a structured graph. The representation consists of four core components: (1) Contexts (abstracted locations), (2) Actions (parameterized affordances), (3) Workflows (observed trajectories), and (4) Tacit Knowledge (domain definitions and reusable procedures). We evaluate this framework on the WebArena benchmark across five…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
