Hosting behaviour
Which region served the request, how long the function took to respond, whether the deployment identifier changes when you expect it to. The diagnostics page reports all of it from the response itself rather than from a dashboard.
Reference property
A small site that proves
the pipeline works.
Usharea is deliberately plain. It exists so that a deployment can be checked end to end - build, region, cache headers, redirects, crawl, index - without a real product getting in the way of the measurement.
Waiting
Edge response
The live status endpoint did not answer. That is expected when the
site is opened from disk or served without the functions in
/api. Everything else on this page is static and
unaffected.
Why it exists
Three things worth checking before anything real ships
Most deployment problems are not application problems. They are routing, headers and indexing problems, and they are far easier to find on a site with nothing else going on.
Crawl and index
Every page is a real document with its own title, description, canonical URL and structured data. There is a sitemap, a robots file and an RSS feed, so a crawler has an unambiguous view of what exists and what it should ignore.
Delivery cost
One stylesheet, one script, no fonts fetched over the network and no analytics. When a page here is slow, the cause is the network path or the host - never four megabytes of framework.
Budget
The whole site, measured
Figures for this page, uncompressed, as committed to the repository. They are checked in the README so a regression is visible in a diff.
1css
Stylesheet
2js
Scripts
0deps
npm packages
0rq
Third-party calls
Under the hood
How it is put together
The panels below are fetched as HTML fragments when you select them. That is the whole htmx idea: the server sends markup, the browser swaps it in, and there is no client-side state to keep in sync.
Plain documents first
Every URL resolves to a complete HTML file. htmx enhances
navigation with hx-boost, but if scripting fails the
links are still links and the forms still post.
No build step
What is in the repository is what is served. There is no bundler, no transpiler and no generated output directory, which removes an entire category of "works locally" problems.
Measure from the response
Region, deployment identifier and timing are read out of real response headers rather than reported by a dashboard, so what you see is what a visitor actually received.
Honest about itself
This is a test property and says so. Nothing here pretends to be a company, a product or a customer, which keeps the indexed content truthful if a crawler does find it.
Notes
Written up as we go
Long-form pages exist mainly to give a crawler something with real structure to chew on: headings, dates, internal links and article markup.
-
Static HTML still wins the boring races
Why a directory of files remains the most predictable thing you can put behind a CDN.
-
Using htmx without hurting your SEO
Partial swaps are fine. Replacing your URLs and your titles is not. Where the line sits.
-
Reading the headers your edge actually sends
Cache status, region codes and request identifiers, and what each one tells you when a deploy looks wrong.
Point a crawler at it
The diagnostics page reports what the host is doing right now. If something looks wrong from where you are sitting, the contact form posts to a serverless function and echoes back what it received.