CubeLoop — a Pythonic, async-native agent harness core designed for high performance, readability, and production-grade persistence.
Python agent-frameworkagent-harnessagent-runtimeagentic-aiai-agentsai-frameworkanthropicharnessharness-engineeringharness-frameworklanggraphlanggraph-alternativellmmcpopenaipi-agentpi-agent-corepythonpython-agent
// Cloudflare Pages "advanced mode" Worker.
//
// The site moved from cubepi.pages.dev → cubepi.ai → cubeloop.dev. The Pages
// project is still named `cubepi`, so the production pages.dev hostname remains
// cubepi.pages.dev (preview URLs are <hash>.cubepi.pages.dev). This 301s that
// hostname — and cubeloop.pages.dev if the project is later renamed — to the
// matching cubeloop.dev path. Search Console Change of Address requires a
// single hop; chaining through cubepi.ai fails Google's 301 check.
//
// Why _worker.js (advanced mode) and not functions/_middleware.js:
// The site is deployed via Direct Upload (cloudflare/pages-action uploads
// website/build). Cloudflare compiles a _worker.js that sits IN the uploaded
// output directory; a functions/ directory, by contrast, must live at the
// project root where Wrangler runs — placing it under static/ (so it lands in
// build/functions/) makes Cloudflare serve it as a static file, not a Function.
// Docusaurus copies static/_worker.js to build/_worker.js, the deployed dir.
//
// Notes:
// - Matches the production hostname exactly, so preview deployments
// (e.g. <hash>.cubepi.pages.dev) fall through and stay browsable.
// - Everything else is forwarded to env.ASSETS.fetch, which serves the static
// site exactly as Pages would (trailing-slash handling, 404.html, etc.).
export default {
async fetch(request, env) {
const url = new URL(request.url);
if (
url.hostname === "cubepi.pages.dev" ||
url.hostname === "cubeloop.pages.dev"
) {
url.hostname = "cubeloop.dev";
url.protocol = "https:";
url.port = "";
return Response.redirect(url.toString(), 301);
}
return env.ASSETS.fetch(request);
},
};