How an Agent Built a 3D Paris Gallery by Chaining Two Hugging Face Spaces

A June 2026 community post on Hugging Face, written by an author who goes by Mishig, describes a coding agent assembling a Paris monument gallery by connecting two hosted model applications on Hugging Face Spaces. One generated monument images; TripoSplat turned each image into a 3D Gaussian splat, a three-dimensional representation a viewer can rotate. The author says every Gradio Space exposes a plain-text agents.md endpoint listing how to call it, with a Hugging Face access credential, so the agent could pass outputs between the two and build the viewing interface. The author reports that the human inputs were taste-level instructions such as zooming out or replacing a monument. The useful pattern is a documented sequence of callable tools: an agent can coordinate specialized models while a person directs the desired result. Gallery build account