From Traces to Agentic Worlds: Agentic Language World Models for Interactive Environment Simulation

arXiv 2026 34.1 method, system

TLDR

Trace2Env uses historical traces to build agentic language world models, enabling stateful environment simulation for LLM agents without executable access.

Reasoning

Strengths include a novel learning-free paradigm, strong empirical validation across nine environments, and improved interaction consistency over prompt-based LWMs. Weaknesses are limited to language-based simulation with no video or dynamic visual modeling, and unknown scalability to complex environments.

Read-first score

Read-first score 34.1, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 42.

Recency 5%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Topical relevance 25%

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Methodology quality 15%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=result

Reproducibility 15%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=reproduce

Graph prestige 15%

5.7

PageRank-style prestige over the local citation graph built from available OpenAlex references. local_pagerank=0.0016

Citation impact 15%

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. cited_by_count=0

Citation velocity 10%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 327.

Keyword Scores

world model

interactive world model

world dynamics prediction

world simulator

generative world model

model-based reinforcement learning world model

video world model

Deep Analysis

Innovations

Proposes agentic language world modeling, where a world model agent serves as the interactive environment for a task agent instead of rebuilding an executable environment.
Introduces Trace2Env, a learning-free framework that reconstructs historical interaction traces into a reusable environment worldbook containing environment schemas, grounded evidence, and induced behavioral knowledge.
Uses persistent episodic state at runtime, with the world model agent consulting the worldbook to infer each action's observation and lasting state effects.
Positions trace-based agentic language world modeling as an alternative direction for building realistic environment replicas without reconstructing the original executable system.

Methodology

Trace2Env reconstructs available historical interaction traces into a reusable worldbook of environment schemas, grounded evidence, and induced behavioral knowledge. At runtime, a world model agent consults this worldbook together with persistent episodic state to infer observations and state effects for a task agent. The framework is evaluated across nine environments against conventional prompt-based language world models using next-observation fidelity, long-horizon interaction consistency, and replay validity of task agent actions.

Key Results

Across nine environments, Trace2Env improves next-observation fidelity and long-horizon interaction consistency over conventional prompt-based language world models. Task agent actions generated against Trace2Env remain valid more often when replayed in the real environment, indicating better preservation of action consequences over multiple turns.

Resources

Paper