Play Doom with computer-use agents as an informal benchmark
The OpenClaw team used computer-use agents to play Doom on remote machines, treating game performance as an informal way to improve the computer-use workflow.
Done withOpenClaw
- What they did
- The source only mentions this in one line of a livestream recap with Vincent Koc. The OpenClaw team has computer-use agents play Doom inside remote machines. Those machines come from their Crabbox remote-execution setup, where a person can open the VNC desktop or SSH in. They treat how well the agents play as an informal benchmark for improving computer-use.
- How it went
- No scores or results are given for the Doom runs. Koc says computer use is now about 10% of his workload, mostly clicking through infrastructure interfaces or verification screens.
- Worth knowing
- The source gives no cost or setup detail for the Doom runs. Crabbox has no built-in cost limits yet, so cap spending on any remote machines you provision yourself.
Try it yourself with OpenClaw
Use the computer to play [Doom or another game] and try to reach [goal, e.g. finish the first level]. Keep a short log of where mouse and keyboard control worked well or badly, and stop after [15 minutes] or when you finish, with a list of what to improve. Ask me before installing software or entering any credentials.
Source: theneuron.ai · Undated
Five of these in your inbox every morning
The best things people got an AI agent to do, each with the prompt to try it.
More like this
Manage 18 agents from a Notion workspace
Notion used as the control plane for a fleet of 18 OpenClaw agents.
Automate freelance admin with a morning briefing agent
An OpenClaw agent sends a Telegram morning briefing, drafts weekly client status emails sent after one-tap approval, and chases follow-ups, cutting admin from about 9 hours to 20 minutes.
Deliver scheduled article digests to Telegram with OpenClaw cron jobs
A developer fixed OpenClaw cron jobs that ran without delivering by turning thinking mode off, after which six articles reached Telegram in 30 seconds.
Play RollerCoaster Tycoon
Ramp Labs put Claude Code in RollerCoaster Tycoon.
- Claude
Let agents play computer games all day
Agents including Claude Opus and o3 were given computers to independently play games like 2048 for hours.
Build a match-3 game from a single prompt
An author prompted Gemini 3 Pro once to produce a working Candy Crush-style game with a shareable link, and says a non-technical friend did too.