Elamur Research⌕
☰
home / papers / 2026 / C-World: A Computer Use Agent Environment Creator

C-World: A Computer Use Agent Environment Creator

Paper · 2026 · Ziqiao Xi, Shuang Liang, Qi Liu et al.

Summary

LLM agents still fall short of humans at long-horizon planning, partly because training environments are few, narrow, and costly to build by hand. C-World addresses this with a system that generates agent environments on demand rather than offering one fixed benchmark. It treats an environment as four parts. The first is a unified library of real MCP tools with a retrieval index. The second is a task generator that produces long, multi-step workflows with "wild" constraints. The third is a state controller that injects failures, timeouts, and state drift. The fourth is a reward scheme mixing verifiable checks with LLM judging. A simulated mode, the World Engine, imitates tool behavior without calling live services, so environments can be scaled up and can even include tools that do not exist yet. In tests on nine frontier LLMs, planning was consistently good, but execution and constraint adherence were weak. Small models fine-tuned on a modest set of C-World trajectories beat baselines trained on far more data. This makes C-World useful both as a testbed and as a data source.

Key points

  • Action space: 5,571 validated tools drawn from 276 MCP servers on Smithery, spanning 204 real applications such as Gmail and GitHub.
    • Each tool passed a three-stage executability check.
    • Agents find tools through a BGE-M3 + FAISS search service.
  • Task length: tasks average 28.5 interaction rounds.
  • Main failure mode: across the nine evaluated LLMs, the biggest problem was following constraints, not calling tools.
    • DeepSeek-V3.2 recovered best from injected disruptions.
    • Making more tool calls did not lead to higher success.
  • World Engine fidelity: the simulated mode, run with gemini-3.1-flash-lite-preview on 50 tasks, ranked models much like real API execution (Spearman ρ = 0.883).
  • Data efficiency: Qwen2.5-7B and Qwen3-8B fine-tuned on 1,170 trajectories outperformed baselines trained on 119k samples.
    • They reached up to 30.05% on BFCL.
    • They stayed robust on MCP-Universe, while the baselines degraded there.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of California San DiegoSan Diego · United States89 works · 54,032 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 works

Cite

Cite as Ziqiao Xi, Shuang Liang, Qi Liu, Jiaqing Zhang, Letian Peng, Fang Nan, Meshal Nayim, Tianhui Zhang, Rishika Mundada, Lianhui Qin, Biwei Huang and Kun Zhou. 2026. C-World: A Computer Use Agent Environment Creator. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.2001

@inproceedings{xi2026c,
  title = {{C-World: A Computer Use Agent Environment Creator}},
  author = {Ziqiao Xi, Shuang Liang, Qi Liu, Jiaqing Zhang, Letian Peng, Fang Nan, Meshal Nayim, Tianhui Zhang, Rishika Mundada, Lianhui Qin, Biwei Huang and Kun Zhou},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.2001},
  url = {https://doi.org/10.18653/v1/2026.acl-long.2001}
}

References

37 references, 21 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Sébastien Bubeck, Varun Chandrasekaran, Ronen Eldan, Johannes Gehrke, Eric Horvitz, Ece Kamar, Peter Lee, Yin Tat Lee, Yuanzhi Li, Scott Lundberg, and 1 others. 2023. Sparks of artificial general intelligence: Early experiments with gpt-4. arXiv preprint arXiv:2303.12712.
  2. Jianlyu Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. 2024. M3-embedding: Multi-linguality, multi-functionality, multi-granularity text embeddings through self-knowledge distillation. In ACL (Findings).
  3. Xiang Fei, Xiawu Zheng, and Hao Feng. 2025. Mcp-zero: Proactive toolchain construction for llm agents from scratch. arXiv preprint arXiv:2506.01056. — [arXiv:2506.01056](https://arxiv.org/abs/2506.01056) find
  4. Zhicheng Guo, Sijie Cheng, Hao Wang, Shihao Liang, Yujia Qin, Peng Li, Zhiyuan Liu, Maosong Sun, and Yang Liu. 2024. Stabletoolbench: Towards stable large-scale benchmarking on tool learning of large language models. In ACL (Findings). find
  5. David Ha and Jürgen Schmidhuber. 2018. Recurrent world models facilitate policy evolution. In Advances in Neural Information Processing Systems (NeurIPS), volume 31. find
  6. Danijar Hafner, Jurgis Pasukonis, Jimmy Ba, and Timothy Lillicrap. 2025. Mastering diverse control tasks through world models. Nature, 640(8059):647–653. find
  7. Wakana Haijima, Kou Nakakubo, Masahiro Suzuki, and Yutaka Matsuo. 2024. The embodied world model based on llm with visual information and prediction-oriented prompts. In ICML 2024 Workshop on Multimodal Foundation Model meets Embodied AI (MFM-EAI). find
  8. Sirui Hong, Mingchen Zhuge, Jonathan Chen, Xiawu Zheng, Yuheng Cheng, Jinlin Wang, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, and 1 others. 2023. MetaGPT: Meta programming for a multi-agent collaborative framework. In The Twelfth International Conference on Learning Representations.
  9. Jeff Johnson, Matthijs Douze, and Herve Jegou. 2021. Billion-scale similarity search with gpus. IEEE Transactions on Big Data, 7(03):535–547. find
  10. Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. 2020. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361.
  11. Guohao Li, Hasan Hammoud, Hani Itani, Dmitrii Khizbullin, and Bernard Ghanem. 2023a. Camel: Communicative agents for" mind" exploration of large language model society. Advances in Neural Information Processing Systems, 36:51991–52008.
  12. Junlong Li, Wenshuo Zhao, Jian Zhao, Weihao Zeng, Haoze Wu, Xiaochen Wang, Rui Ge, Yuxuan Cao, Yuzhen Huang, Wei Liu, and 1 others. 2025. The tool decathlon: Benchmarking language agents for diverse, realistic, and long-horizon task execution. arXiv preprint arXiv:2510.25726.
  13. Minghao Li, Yingxiu Zhao, Bowen Yu, Feifan Song, Hangyu Li, Haiyang Yu, Zhoujun Li, Fei Huang, and Yongbin Li. 2023b. Api-bank: A comprehensive benchmark for tool-augmented llms. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 3102–3116.
  14. Bill Yuchen Lin, Yicheng Fu, Karina Yang, Faeze Brahman, Shiyu Huang, Chandra Bhagavatula, Prithviraj Ammanabrolu, Yejin Choi, and Xiang Ren. 2023. Swiftsage: A generative agent with fast and slow thinking for complex interactive tasks. Advances in Neural Information Processing Systems, 36:23813– 23…
  15. Weiwen Liu, Xu Huang, Xingshan Zeng, Shuai Yu, Dexun Li, Shuai Wang, Weinan Gan, Zhengying Liu, Yuanqing Yu, Zezhong WANG, and 1 others. 2024a. Toolace: Winning the points of llm function calling. In The Thirteenth International Conference on Learning Representations.
  16. Xiao Liu, Hao Yu, Hanchen Zhang, Yifan Xu, Xuanyu Lei, Hanyu Lai, Yu Gu, Hangliang Ding, Kaiwen Men, Kejuan Yang, and 1 others. 2024b. Agentbench: Evaluating llms as agents. In ICLR.
  17. Zhiwei Liu, Jielin Qiu, Shiyu Wang, Jianguo Zhang, Zuxin Liu, Roshan Ram, Haolin Chen, Weiran Yao, Shelby Heinecke, Silvio Savarese, and 1 others. 2025. Mcpeval: Automatic mcp-based deep evaluation for ai agent models. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Pr… find
  18. Ziyang Luo, Zhiqi Shen, Wenzhuo Yang, Zirui Zhao, Prathyusha Jwalapuram, Amrita Saha, Doyen Sahoo, Silvio Savarese, Caiming Xiong, and Junnan Li. 2025. Mcp-universe: Benchmarking large language models with real-world model context protocol servers. In Workshop on Scaling Environments for Agents. find
  19. Grégoire Mialon, Clémentine Fourrier, Thomas Wolf, Yann LeCun, and Thomas Scialom. 2023. Gaia: a benchmark for general ai assistants. In The Twelfth International Conference on Learning Representations.
  20. Guozhao Mo, Wenliang Zhong, Jiawei Chen, Xuanang Chen, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, and Le Sun. 2025. Livemcpbench: Can agents navigate an ocean of mcp tools? arXiv preprint arXiv:2508.01780. — [arXiv:2508.01780](https://arxiv.org/abs/2508.01780) find
  21. Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. 2023. Generative agents: Interactive simulacra of human behavior. In Proceedings of the 36th annual acm symposium on user interface software and technology, pages 1–22. find
  22. Shishir G Patil, Huanzhi Mao, Fanjia Yan, Charlie Cheng-Jie Ji, Vishnu Suresh, Ion Stoica, and Joseph E Gonzalez. The berkeley function calling leaderboard (BFCL): From tool use to agentic evaluation of large language models. In Forty-second International Conference on Machine Learning. find
  23. Shishir G Patil, Tianjun Zhang, Xin Wang, and Joseph E Gonzalez. 2024. Gorilla: Large language model connected with massive apis. Advances in Neural Information Processing Systems, 37:126544–126565.
  24. Yujia Qin, Shihao Liang, Yining Ye, Kunlun Zhu, Lan Yan, Yaxi Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, and 1 others. 2024. Toolllm: Facilitating large language models to master 16000+ real-world apis. In ICLR.
  25. Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. 2023. Reflexion: Language agents with verbal reinforcement learning. Advances in Neural Information Processing Systems, 36:8634–8652.
  26. Guanzhi Wang, Yuqi Xie, Yunfan Jiang, Ajay Mandlekar, Chaowei Xiao, Yuke Zhu, Linxi Fan, and Anima Anandkumar. 2023. Voyager: An open-ended embodied agent with large language models. In Intrinsically-Motivated and Open-Ended Learning Workshop@ NeurIPS2023. find
  27. Ruoyao Wang, Graham Todd, Ziang Xiao, Xingdi Yuan, Marc-Alexandre Côté, Peter Clark, and Peter Jansen. 2024a. Can language models serve as text-based world simulators? In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 1–17, Ban… find
  28. Xingyao Wang, Yangyi Chen, Lifan Yuan, Yizhe Zhang, Yunzhu Li, Hao Peng, and Heng Ji. 2024b. Executable code actions elicit better llm agents. In ICLR 2024 Workshop on Large Language Model (LLM) Agents. on consequential real world tasks. arXiv preprint arXiv:2412.14161.
  29. Zhangchen Xu, Adriana Meza Soria, Shawn Tan, Anurag Roy, Ashish Sunil Agrawal, Radha Poovendran, and Rameswar Panda. 2025. Toucan: Synthesizing 1.5 m tool-agentic data from real-world mcp environments. arXiv preprint arXiv:2510.01179. — [arXiv:2510.01179](https://arxiv.org/abs/2510.01179) find
  30. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. 2022. React: Synergizing reasoning and acting in language models. In The Eleventh International Conference on Learning Representations.
  31. Junjie Ye, Guanyu Li, Songyang Gao, Caishuang Huang, Yilong Wu, Sixian Li, Xiaoran Fan, Shihan Dou, Tao Ji, Qi Zhang, and 1 others. 2025. Tooleyes: Fine-grained evaluation for tool learning capabilities of large language models in real-world scenarios. In COLING.
  32. Shuyan Zhou, Frank F Xu, Hao Zhu, Xuhui Zhou, Robert Lo, Abishek Sridhar, Xianyi Cheng, Tianyue Ou, Yonatan Bisk, Daniel Fried, and 1 others. 2024. Webarena: A realistic web environment for building autonomous agents. In 12th International Conference on Learning Representations, ICLR 2024.
  33. Zhenting Wang, Qi Chang, Hemani Patel, Shashank Biju, Cheng-En Wu, Quan Liu, Aolin Ding, Alireza Rezazadeh, Ankit Shah, Yujia Bao, and 1 others. 2025. Mcp-bench: Benchmarking tool-using llm agents with complex real-world tasks via mcp servers. In Workshop on Scaling Environments for Agents. find
  34. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, and 1 others. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems, 35:24824– 24837.
  35. Zhiheng Xi, Wenxiang Chen, Xin Guo, Wei He, Yiwen Ding, Boyang Hong, Ming Zhang, Junzhe Wang, Senjie Jin, Enyu Zhou, and 1 others. 2025. The rise and potential of large language model based agents: a survey. Sci. China Inf. Sci.
  36. Tianbao Xie, Danyang Zhang, Jixuan Chen, Xiaochuan Li, Siheng Zhao, Ruisheng Cao, Toh J Hua, Zhoujun Cheng, Dongchan Shin, Fangyu Lei, and 1 others. 2024. Osworld: Benchmarking multimodal agents for open-ended tasks in real computer environments. Advances in Neural Information Processing Systems, 37…
  37. Frank F Xu, Yufan Song, Boxuan Li, Yuxuan Tang, Kritanjali Jain, Mengxue Bao, Zora Z Wang, Xuhui Zhou, Zhitong Guo, Murong Cao, and 1 others. 2024. Theagentcompany: benchmarking llm agents find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/xi-2026-c-world-computer-use-agent/