EnvFactory: Scaling Tool-Use Agents via Executable Environments Synthesis and Robust RL

TL;DR

EnvFactory is an automated framework that creates realistic, executable environments and natural trajectories to enhance tool-use capabilities in language models, improving training efficiency and performance.

Contribution

It introduces a fully automated method for synthesizing executable environments and natural trajectories, reducing reliance on costly data and improving agent training.

Findings

EnvFactory achieves up to +15% improvement on BFCLv3.

It generates 2,575 trajectories from only 85 environments.

Outperforms prior methods with fewer environments.

Abstract

Equipping LLMs with tool-use capabilities via Agentic Reinforcement Learning (Agentic RL) is bottlenecked by two challenges: the lack of scalable, robust execution environments and the scarcity of realistic training data that captures implicit human reasoning. Existing approaches depend on costly real-world APIs, hallucination-prone LLM simulators, or synthetic environments that are often single-turn or depend on pre-collected documents. Moreover, synthetic trajectories are frequently over-specified, resembling instruction sequences rather than natural human intents, reducing their effectiveness for RL training. We introduce EnvFactory, a fully automated framework that addresses both challenges. EnvFactory autonomously explores and verifies stateful, executable tool environments from authentic resources, and synthesizes natural multi-turn trajectories through topology-aware sampling and calibrated refinement, producing grounded queries with implicit intents. Using only 85 verified environments across 7 domains, EnvFactory generates 2,575 SFT and RL trajectories. Despite using significantly fewer environments than prior work, which are often 5 times more, EnvFactory achieves superior training efficiency and downstream performance, improving Qwen3-series models by up to +15% on BFCLv3, +8.6% on MCP-Atlas, and +6% on conversational benchmarks including $\tau^2$-Bench and VitaBench. By fully automating both environment construction and trajectory synthesis, EnvFactory provides a scalable, extensible, and robust foundation for Agentic RL.