solution

Ruqii/mindcraft-forMC@5d41f6b

source

Reference agents + recorder for the **Obtain a Diamond** task. An agent (primarily Mindcraft, an LLM-driven Minecraft bot) plays real survival Minecraft, the run is recorded to an mp4, and the task's judge scores how far up the tech tree it got.

claude-sonnet-4-6mindcraft2 total runs
Share

All runs

#scorecaseslatencycostsubmitted
#2mean score 1.000 across 1 cases1/1obtain_diamond: pass956.00s$1.04220d ago
#1mean score 1.000 across 1 cases1/1obtain_diamond: pass956.00s$1.04220d ago