An underground battle royale where you dig through destructible terrain, loot gear, and use darkness, sound, and light to outsmart rivals. Play solo against tactical AI or host browser and macOS WebRTC matches for up to 32 miners.

What became possible in your product with Astra that was not practical before?

GPT-6 Astra made it practical for a solo maker to finish Little Mud as one coherent product instead of a collection of prototypes. Working through Codex, it could hold repo-wide context across the Godot game, WebRTC lobby and signaling, browser and macOS builds, seven-language UI, tactical AI, release scripts, and marketing assets. That let me iterate on gameplay, trace cross-layer multiplayer edge cases, verify exported builds, and produce the launch trailer within the same workflow. Claude Code helped with early prototyping; Astra powered the later production work and release hardening.

Hi Product Hunt — I'm Louis, the solo maker behind Little Mud. Little Mud is a browser-first underground battle royale where the map is part of the fight. You dig routes, loot tools and weapons, choose when to reveal yourself with a headlamp, and survive the gas descending from above. You can jump into a 32-player AI practice match, host a room for friends through WebRTC, or play on macOS. I used Claude Code during early prototyping, then Codex with GPT-6 Astra for the later implementation, multiplayer flow, UI polish, seven-language experience, edge-case debugging, trailer production, and release hardening. The shipped game runs on Godot and does not call an OpenAI API at runtime. I'd especially like feedback on the first five minutes: are the digging, darkness, and sound-based tactics clear, and what should I improve first?

What became possible in your product with Astra that was not practical before?

Little Mud

Dig, hide, and outsmart your friends underground

Dig, hide, and outsmart your friends underground