Aura Open-Sources Autopilot, an AI Agent That Playtests Your Game for Bugs

Most of us only meet game bugs after launch: a door that won't open, a character stuck in a wall, a save file that stops loading. Aura, a team that builds AI helpers for Unreal Engine and Unity developers, announced on October 30 that it is open-sourcing Autopilot, an agent meant to catch more of those problems before players ever see them.

What Autopilot does

Aura describes Autopilot as a self-improving agent that plays your game and finds bugs. It works with Unity, Godot and Unreal, the three engines behind a huge share of indie and mid-size games. The company frames it as computer use for games: a coding agent can already write gameplay code quickly, and Autopilot is meant to close that loop by actually playing the result and reporting what went wrong, so the developer no longer has to test every change by hand.

The claims

According to Aura's announcement, the new Autopilot is:

• 43% more accurate
• almost twice as fast
• better than Claude Code at playtesting

These are Aura's own numbers, and the post doesn't spell out the exact test behind them, so I'd treat them as a starting point rather than a verdict. The code is available on GitHub.

Why it matters to players

Playtesting is one of the slowest, most repetitive jobs in game development, and small teams often can't afford enough of it. An open tool that can work through levels overnight won't replace human testers who notice when something simply feels off, but it could catch obvious, progress-blocking problems earlier. Being open source also means studios can inspect it and run it on their own projects instead of sending early builds to an outside service.

If it works as promised, the next indie game you pick up might ship with a few fewer stuck-in-the-wall moments. Aura's announcement post has a short demo.