wtf is hubproxy?

HubProxy is a self-hosted proxy server that speeds up downloads from Docker Hub, GitHub, and Hugging Face, which is useful in places where direct connections to those services are slow or unreliable. You run it on your own server, point your traffic at it, and it fetches and forwards the files on your behalf, acting as a shortcut in front of the original source. For Docker, it can mirror images from Docker Hub as well as other registries like GHCR, GCR, Quay, and Kubernetes' own registry, so a normal docker pull command can go through your proxy instead of hitting those services directly. It can also export images as offline tar files, either one at a time or in a batch, letting you download a set of images to move onto machines without internet access. For GitHub, it proxies file downloads including release assets, raw files, repository archives, API calls, and even git clone operations. For Hugging Face, it speeds up downloading AI model files. Other features include a search endpoint for looking up Docker Hub images and their available tags and architectures, configurable rate limiting per IP address or IP range, and a wildcard based allow list and block list for controlling which images or GitHub repositories can be proxied. It also exposes a health check endpoint and Prometheus formatted metrics for monitoring. You can run it as a single binary installed with a shell script, as a Docker container, or with Docker Compose. Configuration is set through a TOML file, and nearly every setting can also be overridden with an environment variable. The project builds on an earlier open source project by a different author and is released under the MIT license, meaning it is free to use, modify, and redistribute, including commercially.