Module authoring

Create a module when you want to combine models, change how their results are processed, or add measurements. We recommend starting with an AI assistant, such as ChatGPT or Claude, with tools for editing files and uploading to Hugging Face. Describe what you want to analyze and measure; the assistant builds the module, saves it in your repository, and helps you test it in QuPath.

You do not need this section to use an existing module or train a model with your annotations. A training result is already a module and can be published directly from QuPath.

Before starting, complete one analysis in QuPath and read Models, recipes and modules. You will need a Hugging Face account that can create and edit a model repository.

If you prefer to edit the files yourself, start with Create your first module. The guides below explain recipe.json and Groovy scripts, which QuPath installs and runs as part of a module. QuPath does not provide a recipe editor.

For exact settings, use the Recipe reference, Model compatibility, and Script functions. Adding support for a different neural-network architecture requires changes to HistoForge, not just a new recipe. Contact histoforge.contact@gmail.com to discuss a model contribution.