CrafterDojo: A Suite of Foundation Models
for Building Open-Ended Agents in Crafter

Developing general-purpose embodied agents is a core challenge in AI. Minecraft provides rich complexity and internet-scale data, but its slow speed and engineering overhead make it unsuitable for rapid prototyping. Crafter offers a lightweight alternative that retains key challenges from Minecraft, yet its use has remained limited to narrow tasks due to the absence of foundation models that have driven progress in the Minecraft setting. In this paper, we present CrafterDojo, a suite of foundation models and tools that unlock the Crafter environment as a lightweight, prototyping-friendly, and Minecraft-like testbed for general-purpose embodied agent research. CrafterDojo addresses this by introducing CrafterVPT, CrafterCLIP, and CrafterSteve-1 for behavior priors, vision-language grounding, and instruction following, respectively. In addition, we provide toolkits for generating behavior and caption datasets (CrafterPlay and CrafterCaption), reference agent implementations, benchmark evaluations, and a complete open-source codebase.

CrafterVPT exhibits diverse behavioral patterns across all 22 achievements, plus emergent survival tactics & adaptive strategies

Place Plant → Place Table

Make Wood Sword → Obtain Sapling

CrafterVPT achieves 61.4% Crafter Score, outperforming all baselines methods by up to 29.6%

CrafterCLIP achieves 89.9% R@1 v.s. 1.7% for general-purpose VideoCLIP model

CrafterSteve-1 achieves near-perfect success rates with shorter task completion times across 5 instruction-following tasks

PPO-Steve integrating our foundation models achieves competitive performance across 4 long-horizon tasks

CrafterSteve-1 with instruction chaining from heuristic planner achieves higher success rates across all long-horizon tasks,
highlighting the importance of hierarchical planning for complex, long-horizon tasks in Crafter, like Minecraft-based research

@article{park2025crafterdojo,
  title={{C}rafter{D}ojo: A Suite of Foundation Models for Building Open-Ended Embodied Agents in Crafter},
  author={Park, Junyeong and Cho, Hyeonseo and Ahn, Sungjin},
  journal={arXiv preprint arXiv:2508.13530},
  year={2025}

}