Watch AI agents beat Minecraft's Ender Dragon in under 9 minutes
Made by rmalde
View profile on GitHub (opens in a new tab)This project pairs two AI systems to play Minecraft: a planning model (the creator's own 'Astra') decides the overall strategy, while Jev picks the moment-to-moment actions — movement, camera angle, item use — needed to carry it out, sent through the normal player controls rather than by editing the game's code.
The creator's own verified run beat the game (killing the Ender Dragon and reaching the exit portal) in 8 minutes 43 seconds using 131 Jev decisions and 35 planning calls, with no restarts or manual fixes, a 40% improvement over an earlier 14-minute-31-second run. It's a research/demo codebase, not a packaged app; recordings and full run logs stay on the creator's own machine rather than being hosted.
You could use it to…
- Watch two AI systems split planning and split-second action to beat the game
- See a real verified run kill the Ender Dragon in 8 minutes 43 seconds
- Compare how many Jev decisions versus planning calls a run actually takes
Can I use this?
You'll need Node.js, a local Minecraft Java server, and API keys for the planning and decision models · Setup needed
Worth knowing The timing claims come from the creator's own verified runs and checks, not an independent source; this is a research demo for one speedrun route, not general-purpose Minecraft software, and the full video is kept locally rather than hosted anywhere public. BuildTube has not run or verified this software.
Why it's here
Added 25 September
Jev picks 131 split-second Minecraft moves while a separate model plans the run.
Part of “The Jev wave”, a BuildTube editorial story
New Sept 20 · 546 GitHub stars in 5 days
- 546 people have starred it on GitHub.
Numbers checked on 25 September 2026; not refreshed since.
Behind it
See the code on GitHubBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?