Back to AI game news
AI game news

FROGLINE takes an AI-built frog grappling prototype into a full 3D escape

By GameSlash

Released November 14, FROGLINE uses tongue swings, moving hazards and tree climbing; its creator explains Codex-assisted controls and a packaged-game crash fix.

MivaMagic released FROGLINE for the 2026 Epic AI MegaJam on November 14, 2026. In this 3D platformer, a frog must escape a storm drain by aiming its tongue, swinging between rings, avoiding moving hazards, and climbing a real tree. The developer's dated journal explains how the design progressed beyond a four-ring prototype and how Codex assisted the controls and testing.

Official FROGLINE gameplay screenshot from MivaMagic's itch.io gallery
Original game screenshot by MivaMagic / FROGLINE. This is source gameplay art, not a substitute generated for the article.

A grappling mechanic with a changing final challenge

The early prototype automatically targeted the next ring. In the released build, players aim with the mouse, left-click to cast the tongue, right-click to release, and use the scroll wheel to adjust tongue length by one metre per step. Amber lights highlight rings, while hot pipes break the tether; a water current carries the frog past a rotating fan. The last climb uses actual branches instead of more floating rings, forcing players to reuse free-aiming skills.

Where AI and human testing fit

According to the November 14 developer journal, OpenAI Codex assisted C++ controls, camera, checkpoints and tests; Unreal MCP checked edits inside the engine. ImageGen was used for frog references and source textures, Rodin for 3D generation, and Blender for character work. The creator chose the level flow and personally playtested the builds, rather than claiming autonomous end-to-end creation.

Second distinct official screenshot from the FROGLINE storm-drain platformer
Additional developer image from the public FROGLINE gameplay gallery.

A concrete packaged-build bug involved a spike or fan checkpoint reset. A timer callback still referenced hit-description text after reset had cleared it, causing a crash. The developer reports fixing this by passing the hit cause by value, then repeating both spike and fan tests in the packaged DirectX 12 build. The journal links a 55-second recording from that build. These are creator reports, not independent GameSlash tests.

FROGLINE is listed as a Windows download of approximately 623 MB on the official itch.io page. The development log is a useful case study in turning a single AI-assisted movement mechanic into a readable and complete escape course.

View source