Measuring Frontier Code Agents at Digital-Twin Scene Construction

Simulation-based robot learning is gated by a manual step that repeats at every deployment site: constructing a physical digital twin of the scene. This paper measures how far unmodified, off-the-shelf code agents can automate that step, a setting referred to as Frontier Agent as Scene Constructor (FASC). Frontier code agents such as Claude Code and Codex, each used as shipped, rebuild eight BEHAVIOR-1K rooms from one photograph, an asset catalog and the simulator. Each reconstruction is scored against the room's known 3D layout on ACDC's sim-to-sim protocol, with memorisation and file-access contamination controlled. The strongest agent on the rooms, Claude Code with Opus 5, retrieves the exact asset model for 52% of the scored objects, against 6% for the same agent with Opus 4.6, released seven months earlier. With Opus 5 it also places objects within 39 cm of ground truth across repeated runs, a 4x reduction in placement error over that predecessor. A gap to ACDC's curated pipeline, at 6 cm, remains. Next, the strongest agent extends to outdoor farm scenes, where no calibrated camera or 3D ground truth exists, using a procedural plant model and common asset libraries. Validated against their photographs in DINOv2 embedding space, 75% of the twins rank their own photograph first among the eight reference photographs in cosine similarity. Together with prior results on trajectory generation, these findings indicate that a generic code-agent harness covers a growing share of embodied tasks, and that its capabilities improve with each frontier-model generation while requiring no embodiment-specific harness engineering.