jupyter-loopback transparently proxies kernel-local HTTP/WebSocket servers so they are reachable from the notebook browser in JupyterLab, Notebook, JupyterHub, VS Code, Colab, and other frontends — without per-frontend configuration. It is used by libraries that spin up local web servers inside a kernel (e.g. tile servers, viewers, dashboards) to make their URLs work seamlessly from the browser.