Use Claude to plan, code, and test a game tool
They used Claude Cowork over several months to design a game tool, generate code, and keep refining it through PowerShell, Maya, and screenshots.
- What they did
- A non-coder worked on a game tool for months with Claude Cowork. They kept a Claude.md file as a reference, talked through the system design with Claude, asked it for code, and then executed that code in PowerShell before bringing the output into Maya. They sent screenshots back to Claude for more rounds of changes, and sometimes requested an audit and shared that with ChatGPT for another opinion before moving ahead.
- How it went
- The process was described as workable over a long stretch, but no finished tool or benchmark numbers were reported. One drawback was that Claude could not judge the hair renders directly when only the unshaded spline view was available, so screenshots were needed for review.
- Tips
- Expect to handle the handoff between tools yourself, since the agent was not connected to PowerShell or Maya in one loop. Visual checks mattered, especially when the scene was hard to read without a screenshot.
Try it yourself with Claude Cowork
I’m making a [game/3D] tool in [Maya or another app]. First, help me think through the architecture. Then draft the code one step at a time, and tell me what to run in [PowerShell or another shell] for each step. Stop after every change and wait for my screenshot, log, or error report before making the next edit. If a step would send, submit, delete, or otherwise make an irreversible change, pause and ask for my approval first. Keep iterating until the tool does [target behavior] in my project.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Use paired agents to plan and code a Godot game
One agent handles the game plan and task list while another writes GDScript, checks the build, and prepares commits for approval.
Let an AI play-test your game through a text harness
A developer wrapped a browser game so an AI could run playtests, reproduce bugs, and flag problems before each milestone.
Use Doom runs to test a computer-use agent
Have a computer-use agent play Doom on a remote machine and use its performance to spot workflow problems.
Build a Telegram game-advice bot from screenshots
Use an n8n Telegram workflow with GPT-4o vision to read game images or voice notes and return move suggestions.
Build a strategy game from a Polkadot network layout
A user and Manus spent a day turning Polkadot parachains into a game concept with cross-chain-style events.
Let an agent take turns in a browser game
A writer used ChatGPT agent mode to compete in Ben Orlin’s online math games, stepping in between turns so the agent could continue.