{"paper":{"title":"Agent World Model in the Wild: Computer-Use Agent Environment Synthesis","arxiv_id":"2610.26828","primary_cat":"cs.AI","submitted_at":"2026-10-19T17:55:35+00:00","version_compiled":"v1"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v1","sentences":[{"i":0,"p":1.0,"id":"s0","own":false,"role":"background","text":"Agent learning is shifting from learning from human data toward learning through interaction with the world.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.06,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.21,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":1.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":0.73,"id":"s1","own":false,"role":"background","text":"However, scaling such learning directly in real-world computer systems is challenging and can introduce substantial security risks.","hedge":{"p":0.89,"label":"asserted"},"own_p":0.08,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.05,"role_probabilities":{"gap":0.26,"method":0.0,"result":0.0,"background":0.73,"limitation":0.01,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":2,"p":0.99,"id":"s2","own":true,"role":"contribution","text":"We therefore propose Agent World Model in the Wild (WildAWM), a scalable pipeline for synthesizing controllable environments and tasks for training general computer-use agents (CUAs).","hedge":{"p":0.98,"label":"asserted"},"own_p":0.72,"main_claim_p":0.19,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.0,"method":0.01,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.99,"interpretation":0.0}},{"i":3,"p":0.56,"id":"s3","own":true,"role":"method","text":"Whereas existing approaches typically specialize in a single agent interaction interface, we construct executable environments that jointly support graphical user interfaces, model context protocol services, and command-line interfaces, allowing agents to flexibly combine visual interaction, tool use, and code execution.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.89,"main_claim_p":0.05,"quantitative":false,"quantitative_p":0.28,"role_probabilities":{"gap":0.0,"method":0.56,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.44,"interpretation":0.0}},{"i":4,"p":0.4,"id":"s4","own":true,"role":"method","text":"WildAWM is entirely powered by open-source models, demonstrating that high-quality computer-use environments can be generated without proprietary frontier models.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.89,"main_claim_p":0.08,"quantitative":false,"quantitative_p":0.07,"role_probabilities":{"gap":0.0,"method":0.4,"result":0.06,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.32,"interpretation":0.22}},{"i":5,"p":0.79,"id":"s5","own":true,"role":"method","text":"Beyond environments, we address three recurring challenges in task synthesis: ensuring solvability, maintaining sufficient difficulty, and producing reliable verifiers, by introducing execution-based quality gates that combine heterogeneous agent rollouts, cross-judge consistency checks, and counterfactual probes of judgment robustness.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.94,"main_claim_p":0.06,"quantitative":false,"quantitative_p":0.14,"role_probabilities":{"gap":0.0,"method":0.79,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.21,"interpretation":0.0}},{"i":6,"p":0.99,"id":"s6","own":true,"role":"result","text":"The resulting pipeline synthesizes 341 long-horizon environment-task pairs that can require over 100 agent steps to solve and remain challenging even for frontier models, with GPT-6-Astra solving only 60% of them.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.94,"main_claim_p":0.14,"quantitative":true,"quantitative_p":0.99,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.99,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.01,"interpretation":0.0}},{"i":7,"p":0.89,"id":"s7","own":true,"role":"method","text":"We train two Qwen models of different sizes exclusively in these synthetic worlds.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.96,"main_claim_p":0.05,"quantitative":true,"quantitative_p":0.56,"role_probabilities":{"gap":0.0,"method":0.89,"result":0.01,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.1,"interpretation":0.0}},{"i":8,"p":0.91,"id":"s8","own":true,"role":"result","text":"Both models achieve substantial gains across 8 out-of-distribution benchmarks spanning web, claw, and computer-use agents, demonstrating that learning in WildAWM's synthesized worlds generalizes broadly to unseen computer-use tasks.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.96,"main_claim_p":0.53,"quantitative":true,"quantitative_p":0.96,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.91,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.09}},{"i":9,"p":0.87,"id":"s9","own":true,"role":"contribution","text":"The code is publicly available at https://github.com/MSR-Orchard/WildAWM.","hedge":{"p":0.98,"label":"asserted"},"own_p":0.64,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.0,"method":0.04,"result":0.01,"background":0.01,"limitation":0.0,"future_work":0.07,"contribution":0.87,"interpretation":0.0}}]},"compiled":{"main_claim":{"p":0.99,"text":"Both models achieve substantial gains across 8 out-of-distribution benchmarks spanning web, claw, and computer-use agents, demonstrating that learning in WildAWM's synthesized worlds generalizes broadly to unseen computer-use tasks.","location":"abstract s8","hedge_strength":{"p":1.0,"label":"asserted"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":7668,"cost_usd_reported":0.0},"compiled_at":"2026-10-20","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.08,"states_limitation":0.25,"human_participants":0.05,"controlled_comparison":0.4,"randomized_assignment":0.02,"states_quantitative_result":0.99,"compares_with_prior_results":0.86,"mentions_code_or_data_availability":0.98},"evidential_basis":{"p":0.91,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.0,"observation_or_experiment":0.91,"simulation_or_computation":0.09}},"contribution_type":{"p":0.97,"value":"method or tool","probabilities":{"review":0.0,"method_or_tool":0.97,"empirical_study":0.0,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":0.03,"position_or_commentary":0.0}},"contribution_statement":{"p":0.96,"text":"We therefore propose Agent World Model in the Wild (WildAWM), a scalable pipeline for synthesizing controllable environments and tasks for training general computer-use agents (CUAs).","location":"abstract s2"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v1"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":10,"sentences_in_abstract":10},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448,"review_comments_judged":{"judged_by":["gpt-5.6-luna","grok-4.7"],"judged_against":"Pith's own text copy of each paper: text extracted from the PDF, the first 90,000 characters","measured_on":"2026-10-04","accurate":113,"decided":149,"sampled_reviews":200,"split_by_lane":false,"by_lane":"https://pith.science/accuracy#by-lane"}}}