* Structure in the banner is generated by BuilderGPT with model gemini-2.5-pro.
A generative Minecraft structure tool for the Cynia Agents framework.
BuilderGPT lets you describe a building and have a language model produce a ready-to-use schematic. The component outputs either a .schem file or a .mcfunction script that can be imported into the game with tools like WorldEdit.
We’re so excited to announce that BuilderGPT on ChatGPT is officially available on the ChatGPT GPTs Store. This is a purely non-profit GPT that lets you generate Minecraft building files instantly—no local setup needed.
What’s the difference?
-
BuilderGPT (Local, open-source & free): Built on the CyniaAgents framework. You run it on your own machine, configure the environment, and use your own API key.
-
BuilderGPT on ChatGPT (new): Runs entirely inside ChatGPT and our cloud server. No install, no keys, no config. From today, you can use OpenAI’s GPT-5 model for free to quickly get the files you want!
-
Zero setup—just open and build
-
Fast, consistent outputs powered by GPT-5
-
Free to use, community-first
Try it Here https://chatgpt.com/g/g-68e21df749f48191b94f57c3d5faa57a-mc-builder-buildergpt
- Install the Cynia Agents framework by following the instructions in its repository.
- Copy this
BuilderGPTfolder into the framework'scomponents/directory. - Install the requirements in the same Python environment:
pip install -r components/BuilderGPT/requirements.txt
- Start the Cynia Agents UI (
python -m streamlit run web.py). - Enable BuilderGPT from the Component Center page.
After enabling, a new BuilderGPT page appears in the sidebar. Select your Minecraft version, choose an export format and describe the structure you want. The component will call the configured LLM provider and save the resulting file inside the generated/ folder.
- Generate structures using natural language
- Export as
.schemor.mcfunction - Works entirely within Cynia Agents
- MCBench (mcbench.ai) — BuilderGPT’s support for a JavaScript-based building/structure representation and elements of its prompting approach are inspired by MCBench. The JavaScript parsing and integration code used here was implemented independently for this project.
This project is licensed under the Apache 2.0 License. See LICENSE for details.
