What is animadex?

AnimaDex is a self-hosted web gallery for browsing and searching AI-generated anime character and artist references. You run it on your own machine, and it stores everything locally in a SQLite database with no external services required for the core features. The interface is built with a Python web framework called Flask on the backend and plain JavaScript on the frontend. The gallery is organized around three types of content. Characters get their own pages with filter options for hair color, hair length, eye color, gender, and copyright (the anime series they come from), plus optional links to AI model files on a site called CivitAI. Artists get profiles with optional quality scores generated by an image-scoring model. Copyrights, meaning anime series, get a browse view that shows a small collage of characters from each series. You can populate the gallery in two ways. The first is to sign in at animadex.net, generate an export token, and run the included import script, which downloads the site's full character and artist catalog along with thumbnails. Subsequent runs fetch only new or changed entries. The second way is to build your own dataset from CSV files and images. The README points to a public dataset on Hugging Face that the site itself uses. A built-in pipeline command processes CSV rows one at a time: it adds or updates the database entry, optionally generates the full image using a separate tool called ComfyUI if you have it configured, builds a compressed thumbnail, and scores artists if scoring is enabled. Each step is idempotent, meaning you can re-run the pipeline on existing data without duplicating work. Optional features such as artist scoring, image generation, and CivitAI sync are all disabled by default and can be turned on through a single configuration file. The project is MIT licensed.