arXiv cs.AI· Austin Tudor David Andrews, Liam Wilkinson, Jamie Heagerty, Harry Coppock, Jakob Nicolaus Foerster, Rui Ponte Costa·· 4 小时前AI 评分36
CivBench：基于《文明 VI》的长时程工具型智能体基准
CivBench: A Long-Horizon Benchmark for Tool-Mediated Agents in Civilization VI
AI 导读
研究者发布开源基准 CivBench，通过 MCP 在《文明 VI》中评测大语言模型智能体的长时程能力。单局可超 300 回合、产生数千次工具调用，环境提供 76 个 MCP 工具；团队在 4 个模型家族、23 次运行中发现：智能体每 30-75 回合才查询一次胜利进度（指南要求每 20 回合），且自定规划承诺的执行率 RAG@10 仅为 48.2%-65.8%。环境、日志与指标已开源。
正文 · 原文
Abstract:We present CivBench, an open-source benchmark for evaluating language model agents in long-horizon, tool-mediated environments through the Model Context Protocol (MCP). A single episode can span over 300 turns and produces thousands of tool calls over a large action space, requiring sustained planning, state monitoring, and execution under partial observability. The environment exposes 76 MCP tools and a deterministic narration layer that retrieves and formats game state as structured text. We use CivBench to characterise agent behaviour across four model families in 23 admissible runs. The sample is a pilot, not a model ranking: aggregate outcomes do not reliably discriminate models at this scale. Instead, we introduce two interface-level metrics that the environment makes measurable: Proactive Monitoring Rate (PMR), capturing whether agents actively query latent strategic state, and RAG@10, capturing whether commitments stated in structured planning reflections are executed within ten subsequent turns. Across runs we observe two consistent patterns under a shared playbook protocol. Agents under-monitor strategically relevant state that is available but requires explicit querying: despite playbook guidance to query victory progress every 20 turns, agents do so only every 30-75 turns, and in 8 of 20 detectable defeats they failed to query within the 20-turn warning window before game end. Agents also frequently fail to execute near-term commitments stated in their own planning reflections (RAG@10 between 48.2% and 65.8% across models). Both patterns arise despite tool access and explicit guidance, and we interpret them as deviations under instruction rather than absences of capability. We release the environment, scenarios, logs, metrics, and analysis pipeline at this https URL.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.02459 [cs.AI] |
| (or arXiv:2609.02459v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2609.02459 arXiv-issued DOI via DataCite |
Submission history
From: Austin Andrews [view email]
[v1]
Wed, 2 Sep 2026 11:31:39 UTC (5,132 KB)
[v2]
Wed, 4 Nov 2026 17:39:10 UTC (5,032 KB)
来源：arXiv cs.AI · arxiv.org