# MC-EvoBench: Auditing Experience-Driven Self-Improvement in Minecraft Agents

arXiv: https://arxiv.org/abs/2610.13829 · HTML: https://mlsanity.com/paper/2610.13829/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 27 of its 28 references that are in the map.
- First posted on arXiv 9 Oct 2026.

## Builds on (most cited first, 27 of 27)

- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Voyager: An Open-Ended Embodied Agent with Large Language Models](https://mlsanity.com/paper/2305.16291/) (2023)
- [ExpeL: LLM Agents Are Experiential Learners](https://mlsanity.com/paper/2308.10144/) (2023)
- [MineDojo: Building Open-Ended Embodied Agents with Internet-Scale Knowledge](https://mlsanity.com/paper/2206.08853/) (2022)
- [Video PreTraining (VPT): Learning to Act by Watching Unlabeled Online Videos](https://mlsanity.com/paper/2206.11795/) (2022)
- [MineRL: A Large-Scale Dataset of Minecraft Demonstrations](https://mlsanity.com/paper/1907.13440/) (2019)
- [JARVIS-1: Open-World Multi-task Agents with Memory-Augmented Multimodal Language Models](https://mlsanity.com/paper/2311.05997/) (2023)
- [STEVE-1: A Generative Model for Text-to-Behavior in Minecraft](https://mlsanity.com/paper/2306.00937/) (2023)
- [MineWorld: a Real-Time and Open-Source Interactive World Model on Minecraft](https://mlsanity.com/paper/2504.08388/) (2025)
- [Optimus-1: Hybrid Multimodal Memory Empowered Agents Excel in Long-Horizon Tasks](https://mlsanity.com/paper/2408.03615/) (2024)
- [LifelongAgentBench: Evaluating LLM Agents as Lifelong Learners](https://mlsanity.com/paper/2505.11942/) (2025)
- [Agent World Model: Infinity Synthetic Environments for Agentic Reinforcement Learning](https://mlsanity.com/paper/2602.10090/) (2026)
- [EmbodiedEval: Evaluate Multimodal LLMs as Embodied Agents](https://mlsanity.com/paper/2501.11858/) (2025)
- [Memory Forcing: Spatio-Temporal Memory for Consistent Scene Generation on Minecraft](https://mlsanity.com/paper/2510.03198/) (2025)
- [Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops](https://mlsanity.com/paper/2607.07663/) (2026)
- [MCU: An Evaluation Framework for Open-Ended Game Agents](https://mlsanity.com/paper/2310.08367/) (2023)
- [Optimus-2: Multimodal Minecraft Agent with Goal-Observation-Action Conditioned Policy](https://mlsanity.com/paper/2502.19902/) (2025)
- [Robobench: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models as Embodied Brain](https://mlsanity.com/paper/2510.17801/) (2025)
- [Self-Improvements in Modern Agentic Systems: A Survey](https://mlsanity.com/paper/2607.13104/) (2026)
- [Optimus-3: Dual-Router Aligned Mixture-of-Experts Agent with Dual-Granularity Reasoning-Aware Policy Optimization](https://mlsanity.com/paper/2506.10357/) (2025)
- [EvoAgentBench: Benchmarking Agent Self-Evolution via Ability Transfer](https://mlsanity.com/paper/2607.05202/) (2026)
- [OpenHA: A Series of Open-Source Hierarchical Agentic Models in Minecraft](https://mlsanity.com/paper/2509.13347/) (2025)
- [MrSteve: Instruction-Following Agents in Minecraft with What-Where-When Memory](https://mlsanity.com/paper/2411.06736/) (2024)
- [Embodied Arena: A Comprehensive, Unified, and Evolving Evaluation Platform for Embodied AI](https://mlsanity.com/paper/2509.15273/) (2025)
- [PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents](https://mlsanity.com/paper/2608.04003/) (2026)
- [MinePlanner: A Benchmark for Long-Horizon Planning in Large Minecraft Worlds](https://mlsanity.com/paper/2312.12891/) (2023)
- [Experience Transfer for Multimodal LLM Agents in Minecraft Game](https://mlsanity.com/paper/2604.05533/) (2026)

## Abstract (arXiv)

A central challenge in embodied recursive self-improvement is enabling agents to convert interaction history into capabilities that persist beyond one world and transfer to later tasks. A saved memory or successful final action does not by itself establish that a reusable capability was acquired; evaluation must separate capability change from carried-over resources, unequal source exposure, and early execution failures. We construct MC-EvoBench as a controlled Minecraft benchmark with a fixed catalog of 19 registered evolution chains, including 12 primary evaluation chains and seven calibration or probe chains, together with three graph-aligned paired probe families for transfer. The benchmark represents each task as a task-state graph, defines stage-level environment verifiers, assigns matched source and target roles, resets the target world, and permits only a declared agent artifact to cross the source-target boundary. For each probe, no-prior, related, unrelated, and shuffled source conditions are evaluated under matched environment-step budgets and five seeds; we report TSR, WGP, and VMP and retain provenance records. The benchmark reveals a consistent direct-capability gap between controller and planner-backed variants, while related experience yields no stable transfer: it does not improve either tool probe, its small wood-species gain hinges on one seed, and all intervals include zero. Stage audits further show that early wood-collection and crafting bottlenecks often leave later capabilities unobserved. MC-EvoBench thus offers a controlled testbed for durable, transferable embodied capabilities. Code and data are available at https://github.com/xzw-ustc/mc-bench-open-source