Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments
Terminal-Universe rebuilds executable coding environments from prior agent runs, then uses them to generate new verifiable tasks.
The framework replays file operations from terminal-agent trajectories to recover partial workspaces, then fills in missing files and dependencies with a completion agent. It can reconstruct the original task or synthesize new ones, including cross-workspace queries and multi-round sessions with simulated user feedback. Applied to public trajectories, it produced 37.3k task-sufficient environments. Fine-tuning Qwen3.5-27B on the corpus improved reported benchmark scores on Terminal-Bench 2.1 and EvoCode-Bench v2. HF Daily Papers' note
The framework replays file operations from terminal-agent trajectories to recover partial workspaces, then fills in missing files and dependencies with a completion agent. It can reconstruct the original task or synthesize new ones, including cross-workspace queries and multi-round sessions with simulated user feedback. Applied to public trajectories, it produced 37.3k task-sufficient environments. Fine-tuning Qwen3.5-27B on the corpus improved reported benchmark scores on Terminal-Bench 2.1 and EvoCode-Bench v2. HF Daily Papers' note
score 5