The fleet screen is a terminal. When a team wants to look at the same
thing — or you want the overnight picture on a second monitor — the web
add-on serves it as a page.
Switch it on in tiny → ☰ namespace settings, then:
kubectl port-forward svc/tiny-web 8080
No Ingress ships with it. Reaching the page is exactly the cluster access you already have; there is no new front door and no auth story to get wrong.
The same reading powers tiny diff, so the page
and the CLI cannot disagree.
Its ServiceAccount has get, list and watch — plus pods/exec for
the git reads — and no write verb anywhere.
That is a design decision, not an unfinished feature. Answering a
question in tiny performs the action with the answering human's
credentials, which is what makes the audit log name a
person. A browser button would act as the web server's own account
instead, quietly breaking that. So the page prints
tiny answer q-xxxxx <answer> and lets you run it.