The paper introduces WorkForge, a scalable synthesis framework that builds verifiable work‑agent environments from real‑world resources. By extracting factual anchors from expert workflows and associated files, WorkForge generates task instructions, solution plans, and verifiers that remain traceable to observable workspace evidence. The authors constructed 16.7K environments across 40 professional domains, demonstrating significant performance gains for large language models such as Qwen3.5-35B-A3B-Base and Qwen3.5-27B.
arXiv:2609.39325v1 Announce Type: new Abstract: The ability of Large Language Model (LLM) agents to complete daily and professional work is receiving increasing attention. Training such agents requir...
AgentMercury is a scalable framework that synthesizes executable environments from high‑level business scenarios instead of task‑specific benchmarks. It creates a persistent world with entities, services, tools, and invariants, allowing diverse tasks and interaction trajectories to emerge naturally. The authors generated 4,783 environments across 14 industries and 50 countries, and training reinforcement‑learning agents on them improved performance on enterprise workflows and out‑of‑domain benchmarks, while the construction process itself can be learned to increase authoring success.
The paper introduces TerminalHorizon, an automated engine that synthesizes long‑horizon trajectories for terminal agents by reconstructing environments from real‑world professional work and progressively expanding tasks to require further exploration and improvement. Using this engine, the authors create 1.5K task environments and fine‑tune Qwen3.5‑27B, producing TerminalHorizon‑27B, which shows significant performance gains across 12 benchmarks in coding, agentic production tasks, and reasoning. Case studies demonstrate the model’s ability to experiment beyond initial solutions, refine methods based on feedback, and preserve effective improvements in final deliverables.
arXiv:2610.18067v1 Announce Type: new Abstract: Scaling interactive environments is important for training long-horizon agents through reinforcement learning. These environments often require further...
SkillGym is an automatic pipeline that generates verifiable environments for training skill-use agents. It crawls internet skills, filters for reproducible workflows, and uses a builder‑reviewer process to create difficulty‑controlled tasks with reference solutions and verifiers. The system builds 6.8k environments, collects 19k successful trajectories, and fine‑tunes LLMs from 2B to 122B parameters, improving performance and skill invocation rates.
The paper introduces a self‑supervised framework called software‑in‑the‑loop reconstruction (SWR) that automatically generates reference outputs and verification targets for terminal agents by leveraging existing scientific software workflows. SWR executes multiple input configurations, partitions cases into public observations and hidden evaluations, and uses a hierarchical verifier to assess agent‑generated programs against hidden workflow outputs. The authors demonstrate the approach on 500 workflows across six domains, achieving significant performance gains on the Terminal‑Bench 2 benchmark with the Qwen3.8‑Max and Qwen3.8‑27B models.
arXiv:2608. 10366v1 Announce Type: new Abstract: Real-world data science involves long-horizon workflows that span data wrangling, exploration, modeling, visualization, and validation, and require coordinated use of tools such as notebooks, IDEs, terminals, browsers, and databases within real operating environments.
Interlock is a framework that transforms long‑horizon tasks into typed completion graphs, linking evidence, decisions, and required effects through grounding, coupling, and realization edges. It compiles these graphs into a resettable workspace, a public request, and an evaluation rubric, and uses source‑coverage, public‑input, and agent‑as‑judge reviews to ensure that all mandatory requirements are met. Fine‑tuning Qwen3.6 backbones on Interlock‑generated trajectories improves pass rates on AutomationBench, JobBench, and Harvey LAB, showing that the synthesized experience benefits external task environments.
arXiv:2610.14966v1 Announce Type: new Abstract: Long-horizon agents must complete workflows in which later activities depend on earlier outputs and changes to shared state. A key challenge for evalua...
arXiv:2610.14915v1 Announce Type: new Abstract: Developing terminal-based agents for long-horizon execution demands synthetic tasks that tightly couple complex instructions with authentic, interactiv...
arXiv:2608.23283v1 Announce Type: new Abstract: General-purpose language models can reason and synthesize knowledge, but complex work also requires sustained interaction with files, information sourc...
The paper proposes a new way to describe benchmarks for AI systems that perform knowledge work, outlining four explicit fields: represented activity, tested setting, required work product, and evaluated result. It builds an inventory of 18 work activities from O*NET to enable activity-level reporting across occupations, and evaluates these activities for semantic coherence, algorithm sensitivity, ontology legibility, and human interpretability. The authors apply their framework to three existing benchmarks—GDPval, OfficeQA Pro, and APEX-SWE—to show how different aspects of work can be captured within the same representation.