Featured · In progress
Multi-Agent Experimentation
What do AI agents do when they're given a goal, a set of tools, and then left alone?
A testbed for running AI agents hands-off. Agents on different models (Claude, GPT and a free local model) share one Minecraft world; the model decides and deterministic code acts, and every decision is logged. Minecraft is only the environment: the same loop would drive a codebase or a data pipeline.
What it measures
- Reaching the goal
- Planning
- Recovering from failure
- Working together
- Cost
Agentic AI · Multi-agent systems · LLM evaluation · TypeScript · Node.js · Claude · OpenAI · Ollama · Docker · Mineflayer