# WorkRover: Scaling Long-Horizon Cowork Agent Training Via Recursive Environment Evolution

arXiv: https://arxiv.org/abs/2610.18067 · HTML: https://mlsanity.com/paper/2610.18067/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 31 of its 45 references that are in the map.
- First posted on arXiv 13 Oct 2026.

## Builds on (most cited first, 30 of 31)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding Agents](https://mlsanity.com/paper/2407.18901/) (2024)
- [AgentPRM: Process Reward Models for LLM Agents via Step-Wise Promise and Progress](https://mlsanity.com/paper/2511.08325/) (2025)
- [EnvScaler: Scaling Tool-Interactive Environments for LLM Agent via Programmatic Synthesis](https://mlsanity.com/paper/2601.05808/) (2026)
- [Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence](https://mlsanity.com/paper/2604.18292/) (2026)
- [Workspace-Bench 1.0: Benchmarking AI Agents on Workspace Tasks with Large-Scale File Dependencies](https://mlsanity.com/paper/2605.03596/) (2026)
- [Promoting Efficient Reasoning with Verifiable Stepwise Reward](https://mlsanity.com/paper/2508.10293/) (2025)
- [RLFactory: A Plug-and-Play Reinforcement Learning Post-Training Framework for LLM Multi-Turn Tool-Use](https://mlsanity.com/paper/2509.06980/) (2025)
- [HiPER: Hierarchical Reinforcement Learning with Explicit Credit Assignment for Large Language Model Agents](https://mlsanity.com/paper/2602.16165/) (2026)
- [TRACE: Turn-level Reward Assignment via Credit Estimation for Long-Horizon Agents](https://mlsanity.com/paper/2607.13988/) (2026)
- [ClawGym: A Scalable Framework for Building Effective Claw Agents](https://mlsanity.com/paper/2604.26904/) (2026)
- [ToolForge: A Data Synthesis Pipeline for Multi-Hop Search without Real-World APIs](https://mlsanity.com/paper/2512.16149/) (2025)
- [Hardening Agent Benchmarks with Adversarial Hacker-Fixer Loops](https://mlsanity.com/paper/2606.08960/) (2026)
- [FACET: Preserving Source Intent and Executable State in Terminal Task Synthesis](https://mlsanity.com/paper/2608.18580/) (2026)
- [LocalSearchBench: Benchmarking Agentic Search in Real-World Local Life Services](https://mlsanity.com/paper/2512.07436/) (2025)
- [Why Multi-Step Tool-Use Reinforcement Learning Collapses and How Supervisory Signals Fix It](https://mlsanity.com/paper/2606.26027/) (2026)
- [TRCA: Transition-wise Rubric Credit Assignment for Long-horizon LLM Agents](https://mlsanity.com/paper/2608.16156/) (2026)
- [Anchor: Mitigating Artifact Drift in Agent Benchmark Generation](https://mlsanity.com/paper/2605.26321/) (2026)
- [STAMP: Provenance-Guided Credit Assignment for Deep Search Agents](https://mlsanity.com/paper/2607.11172/) (2026)
- [DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training](https://mlsanity.com/paper/2609.04094/) (2026)
- [RecRM-Bench: Benchmarking Multidimensional Reward Modeling for Agentic Recommender Systems](https://mlsanity.com/paper/2605.11874/) (2026)
- [Meta-Task: Turning Terminal Task Synthesis into a Terminal Task for Scalable Agent Training](https://mlsanity.com/paper/2607.27929/) (2026)
- [Envs-FORGE: Frontier-Optimized Reward-Grounded Environment Synthesis for Agent RL](https://mlsanity.com/paper/2608.14312/) (2026)
- [CoAL-RAG: A Complexity-Aware Legal Retrieval-Augmented Generation Method](https://mlsanity.com/paper/2608.17536/) (2026)
- [TAPO: Tool-Aware Policy Optimization via Credit Transfer for Multimodal Search Agents](https://mlsanity.com/paper/2606.05784/) (2026)
- [Echoverse: Deep, Evolving Environments for Training Computer-Use Agents at Scale](https://mlsanity.com/paper/2607.28074/) (2026)
- [CheckerBench: Can Long-Horizon Agents Synthesize Static-Analysis Checkers?](https://mlsanity.com/paper/2610.07557/) (2026)
- [Efficient Reinforcement Learning for Long-Horizon Tool-Use Agentic Tasks](https://mlsanity.com/paper/2608.10357/) (2026)
- [DuMateBench: Evaluating Autonomous Agents in Complex Real-World Workflows](https://mlsanity.com/paper/2608.26546/) (2026)
- [Coverage, Not Targeting: A Structural Regime in Multi-Turn Agent Credit Assignment](https://mlsanity.com/paper/2609.02417/) (2026)

## Abstract (arXiv)

Scaling interactive environments is important for training long-horizon agents through reinforcement learning. These environments often require further testing and refinement after initial generation, and correcting defects in a task may also require updating its work context and verifier. However, existing construction pipelines offer limited support for such iterative refinement while keeping these components aligned. We introduce WorkRover, a framework for training long-horizon cowork agents through recursive environment evolution. Its harness-driven synthesis enables construction agents to jointly build and validate semantically aligned work contexts, executable tasks, and evidence-grounded verifiers, adapting their actions to workspace state and execution feedback rather than following a fixed construction workflow. To support environment evolution, we introduce Factorial Compatibility Replay (FCR), which uses rollout feedback to evaluate combinations of component revisions through dependency-closed replay and selects the smallest validated repair among tested candidates while preserving task requirements and evaluation constraints. To provide finer-grained training signals, we propose Verifier-to-Turn Credit Assignment (VTCA), which augments trajectory-level GRPO advantages with signals from deterministic workspace-progress checks and assigns a bounded share of positive-progress credit to prerequisite turns using recorded tool-output and artifact dependencies. Using WorkRover, we post-train Qwen3.5-35B-A3B to obtain WorkRover-35B-A3B and evaluate it on several cowork-agent benchmarks. Its reported scores exceed those of Qwen3.5-35B-A3B by 26.27 percentage points on JobBench, 29.32 percentage points on OfficeQA Pro, and 413.25 Elo points on GDPval v2. These results suggest that WorkRover offers a promising foundation for scalable training of long-horizon cowork agents.